ACE-FTS Version 3 . 0 Data Set : Validation and Data Processing Update

This Article is brought to you for free and open access by the Chemistry & Biochemistry at ODU Digital Commons. It has been accepted for inclusion in Chemistry & Biochemistry Faculty Publications by an authorized administrator of ODU Digital Commons. For more information, please contact digitalcommons@odu.edu. Repository Citation Waymark, Claire; Walker, Kaley A.; Boone, Chris D.; and Bernath, Peter F., "ACE-FTS Version 3.0 Data Set: Validation and Data Processing Update" (2013). Chemistry & Biochemistry Faculty Publications. 37. https://digitalcommons.odu.edu/chemistry_fac_pubs/37


II. VALIDATION RESULTS
An extensive validation exercise was undertaken for the ACE-FTS baseline species (VMR profiles of O 3 , H 2 O, CH 4 , N 2 O, NO 2 , NO, HNO 3 , HCl, HF, CO, CCl 3 F, CCl 2 F 2 , N 2 O 5 , and ClONO 2 ) and temperature for the version 2.2 (+updates for O 3 , N 2 O 5 and HDO) data set.The retrievals for this version have been described in detail in Boone et al. (2005).The validation results were reported in a special issue of Atmos.Chem.Phys.(http://www.atmoschem-phys.net/special_issue114.html).By building on these comparison results, a newer version of the ACE-FTS data set (version 3.0) was produced that incorporated a new set of microwindows and updated spectroscopic parameters.
It addressed the unphysical oscillations that were found in the mesospheric temperature profiles and an artefact (or "glitch") that occurred in the temperature profiles near 23 km.In addition, the altitude ranges for nearly all species have been extended.These increases range from a few km to as much as 35 km for N 2 O.These improvements have been briefly summarized in Table 1 and are documented in Boone et al. (2013).For the version 3.0 data set, work has been undertaken to characterize the results by performing both comparisons between versions and comparisons with measurements by other satellite instruments.Herein, we focus on describing the direct version comparisons for the O ACE-FTS baseline species.These comparisons have been carried out in order to identify the changes to the ACE-FTS retrievals between the new version 3.0 and the previous wellvalidated (version 2.2 +updates) data set.The-These results allow users to understand the data quality of the version 3.0 data set in relation to the earlier validation studies for the version 2.2+updates data set.Temperature interpolation in P/T retrieval Altitude interpolation approach for temperature was changed to match what was used for VMR retrieval.This fixed the unphysical oscillations observed in v2.2 temperature profiles.

Issue at ~23 km
No empirical function is used in the retrieval of pressure below 23 km in v3.0; pressure at each analysed measurement is used as a fitting parameter Altitude lower limit in P/T retrieval Changed from 12 km (v2.2) to 15 km (v3.0)

High altitude retrieval in P/T retrieval
The retrieved CO 2 VMR profile at high altitudes was forced to match fixed CO 2 VMR at the interface (near 60 km) Tangent height separation in P/T retrieval The tangent height separation is calculated in a way that improves the stability of the retrieval compared to v2.2 The comparison approach taken for this work follows the method described by Dupuy et al. (2009).Rather than finding pairs of "coincident" measurements by different instruments, the pairs used here are all ACE-FTS occultations for which profiles are available for both data versions.For each trace gas species (or subset of these data), the mean profiles and 1-σ standard deviation of the mean are calculated for each data version.The absolute and relative differences are calculated from individual pairs of profiles and then the means of these differences are calculated (see Eqs. 3 and 4, respectively, in Dupuy et al., 2009).To calculate the relative differences, the mean of the version 2.2+updates and version 3.0 VMR is used as the denominator.As was done for most of the version 2.2+updates validation studies, the standard deviation of the bias-corrected differences (which will be referred to as the "debiased standard deviation") is calculated for these comparisons (e.g.Eq. 5 in Dupuy et al., 2009).This gives a measure of the combined precision of the two ACE-FTS data versions (von Clarmann 2006).For the results shown here, no data screening has been applied in order to examine all of the profiles produced for the two ACE-FTS data versions.
The ACE-FTS measurements made between 21 February 2004 (beginning of routine operations) and 30 September 2010 have been used in this work.This analysis has been performed for the full latitude range as well as for selected latitude bands (typically using 30º bins) for all fourteen of the ACE-FTS baseline species.An example of the direct version comparisons for O 3 is shown in Figure 1 for the latitude band between 30º S and 60º S. The results shown here are consistent with those seen in all of the different latitude bands.Figure 2 shows two additional examples of these direct version comparisons for CH 4 and H 2 O.
In addition to subdividing the data into various latitude bands, the data have also been compared separately for different time periods.Figure 3 shows an example for O 3 where the relative differences for each year are calculated separately to examine the year-to-year changes.It can be seen that these differences are quite consistent for most years, with all years seeing quite large variability below ~20 km.However, above ~20 km, 2008 has larger variability than the other years as shown in the de-biased standard deviation of the mean difference.This is primarily due to increased numbers of outlier profiles, which we are currently in the process of characterizing to understand their origin and to provide guidance to users to allow these to be filtered out as needed.A list of occulations with known issues is provided by the ACE Science Operations Centre (https://databace.scisat.ca/validation/data_issues.php).In addition to consulting this list, users are encouraged to submit reports the ACE team outlining any issues they find.A summary of the results obtained for the direct version comparisons of the ACE-FTS baseline species is shown in Table 2.For each species, the reference describing the version 2.2+updates validation study and the altitude range for the version 3.0 retrievals are listed.

III. ACE-FTS PROCESSING PLANS
The next processing version (version 4.0) is now in development (Boone et al., 2013).The primary motivation for this update is to develop a data set that is more appropriate for studying longer-term changes and investigating trends.In ACE-FTS processing versions 3.0 and earlier, the assumed rate of change in CO 2 as a function of time is too low and will therefore be changed to match better with observations.In version 4.0, the shape of the CO 2 VMR profiles at low altitudes will vary with latitude, and a seasonal cycle will be included, features that were not present in previous processing versions.
An issue was identified in the input a priori temperature/pressure profiles for low altitudes that has affected all ACE-FTS retrievals beginning in October 2010 (Boone et al., 2013).Because of this, data from both version 2.2+updates and version 3.0 should not be used after 30 September 2010.New processing versions (2.5 and 3.5) are being produced to provide corrected results for the affected time period (October 2010 onward).These new versions do not include any changes in the retrieval process other than employing more appropriate a priori pressure and temperature information (Boone et al., 2013).
IV. CONCLUSION ACE-FTS and the SCISAT satellite continue to perform well in their tenth year in orbit and produce a valuable data set for investigating the composition of the Earth's atmosphere.Through direct profile comparisons, the current ACE-FTS v3.0 data set is generally seen to improve on the v2.2+updates data set.Comparisons with other satellite data sets are in progress and these results are being used to provide feedback for future ACE-FTS data versions.

Figure 1 :
Figure 1: An example of the ACE-FTS direct version comparisons for O 3 for the 30-60º S latitude range.This compares the version 2.2 O 3 update (blue) product with the version 3.0 (red) product.The mean VMR profiles (solid lines) and the 1-σ standard deviations (dot-dashed lines) are shown in the left panel, the absolute differences are shown in the centre panel and the relative differences are shown in the right panel.The dashed lines in the centre and right panels indicate the de-biased standard deviation of the mean differences.

Figure 2 :
Figure 2: Two further examples of version 2.2 versus version 3.0 comparisons (mean profiles and absolute differences) for CH 4 (left panels) and H 2 O (right panels).The legend for each pair of panels is the same as given for the corresponding panel in Fig. 1.Measurements from all latitudes are included in these comparisons.

Figure 3 :
Figure 3: The mean relative differences for the direct version comparisons for O 3 are shown for each year from 2004 (far left) to 2010 (far right).

Table 2 :
Summary of the version 2.2+updates and version 3.0 direct version comparisons.