Atmospheric Chemistry Experiment
Science Operations Center
Department of Chemistry
University of Waterloo
Waterloo, Ontario, N2L 3G1
ACE – FTS
Atmospheric Chemistry Experiment
Data usage guide and file format description for
ACE-FTS level 2 data version 4.1 ASCII format
Document Number: ACE-SOC-0036
Issue:
1
Revision:
Issue Date: 26 June 2020
Function
Prepared by:
Project Scientist
Approved by: Mission Scientist
Name
Chris Boone
Peter Bernath
Signature
Date
26 Jun 2020
Table of Contents
1. Introduction, Data Usage Information and Format Description ................................................. 4
General comments on the file formatting and data usage:........................................................................ 4
Version 4.1 release dataset: ...................................................................................................................... 6
Use of a priori data: ................................................................................................................................ 10
Use of CO2 profiles provided in data files: ............................................................................................. 10
Validation results for ACE-FTS: ............................................................................................................ 10
Availability of Geolocation Information for ACE profiles: ................................................................... 10
Recommendations for data use and error reporting:............................................................................... 11
Recommendations for quality checking and filtering:............................................................................ 11
A few words on ACE-FTS altitude resolution: ...................................................................................... 14
2.
Readme Files ......................................................................................................................... 17
Index of Tables
Table 1:
Table 2:
Table 3:
Table 4:
Table 5:
Table 6:
File naming convention ………………………………………………………………...5
fields included in netCDF files …………………………………………………………6
Flag values associated with ACE-FTS level 2 data for CO2 .…….………………….....6
File format for “1 km” grid data files – Version 4.1 for main isotopologues…………..7
File format for "1 km" grid data files – Version 4.1 for subsidiary isotopologues……..8
File format for "1 km" grid geolocation (GLC) files …………………………………..9
Index of Figures
Figure 1:
Figure 2:
Figure 3:
Figure 4:
Figure 5:
Figure 6:
All SO2 results from sunsets in May 2012 plus two occultations in 2015…...……….12
All SO2 results from sunsets in May 2012 above 10 km………………………...……13
Average SO2 profiles with and without negative values included………….…..….....14
Width of the ACE-FTS field of view as a function of distance from center …..……..15
ACE-FTS weighting functions for measurements with 2 km spacing………………..15
ACE-FTS weighting functions for measurements with 1 km spacing ...……………..16
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 2
DOCUMENT CHANGE RECORD
Issue
Rev.
1
-
Date
26 Jun 2020
Change Detail
First Issue of document – for version 4.1 data release
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 3
1. Introduction, Data Usage Information and Format
Description
The user is strongly encouraged to read the README files (produced by Chris Boone for each
version of the ACE-FTS retrievals and reproduced in this document) and the ACE-FTS retrieval
paper (C. D. Boone et al., “Retrievals for the atmospheric chemistry experiment Fourier-transform
spectrometer”, Appl. Opt., 44(33), 7218-7231, (2005) available from the ACE Mission website
(http://www.ace.uwaterloo.ca/publications.html)) for more information on the ACE-FTS
retrievals. Also read the ACE-FTS version 3.5 retrieval paper, found on the same website: C. D.
Boone et al. “Version 3 Retrievals for the Atmospheric Chemistry Experiment Fourier Transform
Spectrometer (ACE-FTS)” in The Atmospheric Chemistry Experiment ACE at 10: A Solar
Occultation Anthology (Peter F. Bernath, editor, A. Deepak Publishing, Hampton, Virginia, U.S.A.,
2013). The paper describing version 4.0 and 4.1 retrievals is also available on the web site: C.D.
Boone, et al., Version 4 retrievals for the atmospheric chemistry experiment Fourier transform
spectrometer (ACE-FTS) and imagers, Journal of Quantitative Spectroscopy and Radiative
Transfer, 247, 106939, 2020; doi: 10.1016/j.jqsrt.2020.106939.
General comments on the file formatting and data usage:
In all of these files, the start and end times given (either time stamp or date and time) correspond
essentially to the start and end of the command sequence. They cannot and should not be used to
derive the length of an occultation since they include warm up time and calibration measurements
(deep space and exo-atmospheric). The location given for each occultation is obtained from the
latitude, longitude and time of the 30 km tangent point (calculated geometrically).
A fill value of -999 is used when at each altitude where a retrieval is not performed. The user
should be careful to distinguish fill values (-999) reported in the VMR statistical error columns
from flagged values (-888). This is not a typographical error! For VMR retrievals, the profile
above the highest analyzed measurement is taken as a constant times the input guess profile. These
data are flagged with an error of -888 and are not intended for use in scientific analysis. They are
provided for the sake of completeness.
It should be noted that there are no errors provided for the temperature or pressure retrievals
because of amount of time required to calculate them. The user is directed to the retrieval papers
described above for details on how temperature and pressure retrievals are done. The accuracy of
these retrievals was assessed for version 2.2 and is reported in the validation paper by R. J. Sica et
al. (“Validation of the Atmospheric Chemistry Experiment (ACE) version 2.2 temperature using
ground based and space-borne measurements”, Atmos. Chem. Phys., 8, 35-62, 2008; available from
the ACE website using the link given above).
The ACE-FTS measurements are recorded every 2 s. This corresponds to a measurement spacing
of 2-6 km which decreases at lower altitudes due to refraction. The typical altitude spacing
changes with the orbital beta angle, which is the angle between the orbital plane and the Sun-Earth
vector. Forward model calculations are carried out on a standard 1 km altitude grid (every 1 km
from 0.5 to 149.5 km). During the least squares fitting, values from the retrieval grid (typically
defined by the measurement tangent heights unless the tangent height separations get too small:
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 4
see the readme for version 4.0) are interpolated onto the 1 km grid using a piecewise quadratic
method. For ACE-FTS version 1.0, the results were reported only on the 1 km grid. For versions
2.0 through 4.0, both the retrieval grid and the 1 km grid profiles are available. For version 4.0,
results on the “retrieval grid” may not coincide with the actual altitudes of the fitted variables for
occultations that have small tangent height spacings, because different molecules could have
slightly different retrieval grids. For such occultations, the reported results have been interpolated
onto the “nominal” retrieval grid (i.e., the retrieval grid employed in previous processing versions
for the occultation). For version 4.1, an issue with outputting data into the retrieval grid files
means that these files will not be provided. The data will only be available on the 1 km grid.
This version includes the retrieval of subsidiary isotopologues from the ACE-FTS spectra. The
line strengths in the HITRAN database are scaled according to natural isotopic abundance. For the
subsidiary isotopologues, in order to obtain the actual VMR values, the user will need to scale the
retrieved profile with the isotopic abundances assumed by HITRAN (http://www.hitran.com). We
have adopted the same notation as used by HITRAN to label the subsidiary isotopologues. For
example, the minor water vapor isotopologues are labeled 181 for H218O, 171 for H217O, and 162
for HD16O.
For each occultation event, two ASCII formatted data files are produced: the 1 km grid results for
the most abundant or main isotopologues and the 1 km grid results for the subsidiary
isotopologues. The file naming conventions for the files are given in Table 1. The file formats
are provided in Table 4 (main isotopologues) and Table 5 (subsidiary isotopologues). In addition,
netCDF files containing all profiles (on the 1 km altitude grid) are produced for each species. The
file format for the netCDF files is provided in Table 2.
Table 1: File naming convention
Name
Isotopologues
Data file type
sxXXXXXv4.1.asc
Most abundant
1 km altitude grid
sxXXXXXv4.1iso.asc
Subsidiary
1 km altitude grid
where: sx is the type of occultation, sunrise (sr) or sunset (ss),
and XXXX (or XXXXa or XXXXX or XXXXXa) is the orbit number
The occultation labels are generated using the satellite orbit number calculated by a program called
Systems Tool Kit (STK). For example, ss12345 would be the sunset observed on the satellite's
12345th orbit. On occasion, STK glitches when counting the number of orbits, and two
occultations end up getting the same name. Because we are constrained to use unique identifiers
in occultation names, a number of measured occultations wound up not being processed because
an existing occultation with the name it 'wanted' already existed.
To recover these missed occultations (and avoid renaming thousands of occultations), an 'a' was
appended to the occultation label wherever an occultation of the same name already existed, in
order to generate a unique identifier. Occultations with the same orbit number in the label (e.g.,
ss12345 and ss12345a) are completely different occultations, collected at different times and
locations that just happened to be assigned the same orbit number through a glitch in STK
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 5
calculations. The measurement time and location reported in the file header is the important
information for making use of these data.
Version 4.1 release dataset
Table 2: fields included in netCDF files.
Parameter
altitude
sunset_sunrise
orbit
year
month
day
hour
latitude
longitude
beta_angle
species
species_error
Definition
Tangent altitude for retrieved species, temperature and pressure (in km)
Type of occultation flag (ss = 0 or 2; sr = 1 or 3)
Orbit number for occultation
Year of occultation 30 km geometric tangent point (YYYY in UTC)
Month of occultation 30 km geometric tangent point (MM in UTC)
Day of occultation 30 km geometric tangent point (DD in UTC)
Time of occultation 30 km geometric tangent point (hh.xxxx in UTC)
Latitude of 30 km geometric tangent point (in degrees; ±90, N = +, S = -)
Longitude of 30 km geometric tangent point (in degrees; ±180, E = +, W = -)
Beta angle of occultation at 30 km tangent point (in degrees)
Retrieved volume mixing ratio for species (in ppv; parts per volume)
Statistical error for species retrieval from fitting (in ppv; if this value is -888,
vmr is not retrieved. It is the value obtained by scaling the a priori value)
temperature
Temperature (in K)
temperature_fit
Values indicating if temperature was retrieved from data (1) or is set to the a
priori value (0)
pressure
Pressure (in atm; 1 atm = 1.01325 bar)
CO2_flag*
Flag for the CO2 result (as described in Table 3)
*Note: The parameter CO2_flag is only included in the file for CO2 isotopologue 1 (16O12C16O).
The files for all other isotopologues will have one fewer field.
Table 3: Flag values associated with ACE-FTS level 2 data for CO2.
Flag
value
1
2
3
4
5
Definition
The altitude region where CO2 VMR was fixed in pressure/temperature analysis
(typically 18 to ~60 km). Not for scientific analysis.
Altitude region where CO2 VMR was fitted using an empirical function during
pressure/temperature analysis (typically ~60 to ~125 km). Available for scientific
analysis.
Above highest analyzed measurement (> ~125 km). Not for scientific analysis.
Tangent heights in this altitude region determined from the N2 continuum (typically
5 to 18 km). Available for scientific analysis
Below the lowest analyzed measurement, no data available.
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 6
Table 4: File format for “1 km” grid data files – Version 4.1 for main isotopologues
Filename for “1 km” grid data: sxXXXXXv4.1.asc
Field name
Description
Acceptable values / units
Type
Header section
Name
Occultation identifier using mission
ace.sxXXXX or ace.sxXXXXX or
String
name (ace), orbit number (XXXX or
ace.sxXXXXa or ace.sxXXXXXa
XXXXa or XXXXX or XXXXXa)
and occultation type (sx = ‘sr’ or ‘ss’)
start_timetag
Time stamp for start of measurement
Mission elapsed seconds
Float
sequence for the occultation
end_timetag
Time stamp for end of measurement
Mission elapsed seconds
Float
sequence for the occultation
start_time
Start date and time of occultation
YYYY-MM-DD hh:mm:ss.ms+00
String
measurement sequence (UTC)
end_time
End date and time of occultation
YYYY-MM-DD hh:mm:ss.ms+00
String
measurement sequence (UTC)
Date
Date and time of occultation 30 km
YYYY-MM-DD hh:mm:ss.ms+00
String
geometric tangent point (UTC)
Latitude
Latitude of 30 km geometric tangent
Degrees (±90, N = +, S = -)
Float
point for occultation
Longitude
Longitude of 30 km geometric tangent Degrees (±180, E = +, W = -)
Float
point for occultation
beta_angle
Beta angle of occultation (at 30 km
Degrees
Float
tangent point)
Data section
Z
Tangent altitude grid for retrieved
km
Float
parameters and species
T
Temperature
K
Float
T_fit
Values indicating if temperature was
0 (not fit),
Integer
retrieved from data (1) or is set to the
1 (fit)
a priori value (0)
P (atm)
Pressure
atm (1 atm = 1.01325 bar)
Float
Dens
Atmospheric number density
molecule cm-3
Float
Species
Retrieved volume mixing ratio for
ppv (parts per volume)
Float
species
NOT ppm or ppb
species_err
Statistical error for species retrieval
ppv
Float
from fitting (if this value is -888, the
vmr is not retrieved. It is the value
obtained by scaling the a priori value)
CO2_flag
Flag for the CO2 result
See Table 3
Species and statistical errors are provided in the following order
H2O, O3, N2O, CO, CH4, NO, NO2, HNO3, HF, HCl, OCS, N2O5, ClONO2, HCN, CH3Cl, CF4, CCl2F2,
CCl3F, COF2, COCl2, COClF, C2H6, C2H2, CHF2Cl, HCOOH, SF6, HO2NO2, H2O2, H2CO, CH3OH, CCl4,
N2, O2, CFC113, HCFC141b, HCFC142b, HFC134a, PAN, CHF3, acetone, SO2, CH3CN,ClO, and CO2
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 7
Table 5: File format for “1 km” grid data files – Version 4.1 for subsidiary isotopologues
Filename for “1 km” grid data: sxXXXXXv4.1iso.asc
Field name
Description
Acceptable values / units
Type
Header section
Name
Occultation identifier using mission
ace.sxXXXX or ace.sxXXXXa or
String
name (ace), orbit number (XXXX or
ace.sxXXXXX or ace.sx.XXXXXa
XXXXa or XXXXX or XXXXXa)
and occultation type (sx = ‘sr’ or ‘ss’)
start_timetag
Time stamp for start of measurement
Mission elapsed seconds
Float
sequence for the occultation
end_timetag
Time stamp for end of measurement
Mission elapsed seconds
Float
sequence for the occultation
start_time
Start date and time of occultation
YYYY-MM-DD hh:mm:ss.ms+00
String
measurement sequence (UTC)
end_time
End date and time of occultation
YYYY-MM-DD hh:mm:ss.ms+00
String
measurement sequence (UTC)
Date
Date and time of occultation 30 km
YYYY-MM-DD hh:mm:ss.ms+00
String
geometric tangent point (UTC)
Latitude
Latitude of 30 km geometric tangent
Degrees (±90, N = +, S = -)
Float
point for occultation
Longitude
Longitude of 30 km geometric tangent Degrees (±180, E = +, W = -)
Float
point for occultation
beta_angle
Beta angle of occultation (at 30 km
Degrees
Float
tangent point)
Data section
Z
Tangent altitude grid for retrieved
km
Float
parameters and species
T
Temperature
K
Float
T_fit
Values indicating if temperature was
0 (not fit),
Integer
retrieved from data (1) or is set to the
1 (fit)
a priori value (0)
P (atm)
Pressure
atm (1 atm = 1.01325 bar)
Float
-3
Dens
Atmospheric number density
molecule cm
Float
Species
Retrieved volume mixing ratio for
ppv (parts per volume)
Float
species
NOT ppm or ppb
species_err
Statistical error for species retrieval
Ppv
Float
from fitting (if this value is -888, the
vmr is not retrieved. It is the value
obtained by scaling the a priori value)
Species and statistical errors are provided in the following order
H2O (181), H2O (171), H2O (162), H2O (182), CO2 (636), CO2 (628), CO2 (627), CO2 (638), CO2 (637),
O3 (668), O3 (686), O3 (667), O3 (676), N2O (456), N2O (546), N2O (448), N2O (447), CO (36), CO (28),
CO (27), CO (38), CH4 (311), CH4 (212), OCS (624), OCS (632), OCS (623), NO2 (656), and HNO3 (156)
Note: no results currently provided for H2O (182), CO2 (637), CO (38), or OCS (623)
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 8
Table 6: File format for “1 km” grid geolocation (GLC) files.
Filename for “1 km” grid GLC files: ace.sxXXXXX.txt
Field name
Description
Acceptable values / units
Header section
!OCCULTATION_NAME Occultation identifier using
ace.sxXXXX, ace.sxXXXXa,
mission name (ace), orbit number
ace.sxXXXXX, ace.sxXXXXXa
(XXXX or XXXXa or XXXXX or
XXXXXa) and occultation type (sx
= ‘sr’ or ‘ss’)
!SCALAR_DATE
Date and time of occultation 30 km YYYY-MM-DD
geometric tangent point (UTC)
hh:mm:ss.ms+00
!SCALAR_LATITUDE
Latitude of 30 km geometric
Degrees (±90, N = +, S = -)
tangent point for occultation
!SCALAR_LONGITUDE Longitude of 30 km geometric
Degrees (±180, E = +, W = -)
tangent point for occultation
!BETA_ANGLE
Beta angle of occultation (at 30 km Degrees
tangent point)
!COLUMNS
Definition of columns included in
z p T Lat Lon SunHeading
file altitude (z), pressure (p),
temperature (T), latitude (Lat),
longitude (Lon), sun heading
(SunHeading)
!FORMAT_STRING
Fortran format string for columns
%12.5 for each of the six fields
Data section (each line)
Z
Tangent altitude for retrieved
km
parameters and species
P
Pressure at tangent altitude
atm (1 atm = 1.01325 bar)
T
Temperature at tangent altitude
K
Lat
Latitude of tangent altitude
Degrees (±90, N = +, S = -)
Lon
Longitude of tangent altitude
Degrees (±180, E = +, W = -)
SunHeading
Angle from the ACE-sun pointing
Degrees
vector to the meridional direction
NOTE: The SunHeading
(vector tangent to a longitude
parameter calculation is not
circle)
fully implemented in the v3.5
GLCs and should not be used
for analyses.
• In files for “1 km” grid data, z provided from 0.5 to 149.5 km
When refraction model calculation failed, only the first four lines are provided in the file.
Type
String
String
Float
Float
Float
String
String
Float
Float
Float
Float
Float
Float
Use of a priori data:
For the temperature and pressure profiles, the reported values come from different sources
depending on the altitude range. Below 18 km, these are fixed to meteorological data from the
CMC (Canadian Meteorological Center). Between 18 km and ~125 km, pressure and temperature
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 9
are retrieved. Above ~125 km, the shapes of the pressure and temperature profiles are taken from
MSIS (NRLMSISE-00) model calculations, and the two profiles are each scaled to avoid
discontinuities with the retrieved results below 125 km during the pressure/temperature analysis.
The regions employing a priori data are identified by the T_fit parameter. If the parameter is true
(“T” in version 1.0 or “1” in subsequent processing versions), then the temperature and pressure
have been retrieved from the measurements.
It should be noted that the only places that we use a priori profiles are the areas described above:
p/T is fixed to a priori below 18 km and above ~125 km, and the VMR above the highest analyzed
measurement for a given molecule is taken as a constant times the a priori profile (in this case,
only the shape of the a priori profile is important). The only exception is CO2, where the VMR
profile for CO2 above 125 km is pre-retrieved prior to performing the pressure/temperature
analysis. A simple linear variation with altitude is assumed for the pre-retrieved high altitude CO2
VMR profile. Beyond the exceptions described above, a priori profiles are used only as a first
guess in the least squares fitting. The analysis does not employ optimal estimation.
Use of CO2 profiles provided in data files:
Care should be taken in use of the CO2 profiles provided in these data files. Above ~60 km, CO2
VMR is retrieved from the ACE-FTS spectra and can be used for scientific studies. For altitudes
between 18 and ~60 km, CO2 VMR is held fixed during the pressure/temperature retrieval process,
and the retrieved CO2 in that altitude region will therefore simply reproduce the input assumptions
for the molecule's VMR profile. CO2 VMRs in this altitude region are not to be used for
scientific studies.
New in versions 4.0 and 4.1, because tangent heights below 18 km are determined from
the N2 continuum in ACE-FTS spectra, we can now perform an independent retrieval at low
altitude, between 5 and 18 km. These data may be used for scientific studies. Flags in the data
file indicate the nature of CO2 at the given altitude, as described in Table 4.
Validation results for ACE-FTS:
The ACE mission website contains links to all of the papers produced using ACE data and current
and previous version validation results: http://ace.uwaterloo.ca/publications.html.
Availability of Geolocation Information for ACE profiles:
With this release of ACE-FTS data, geolocation information (GLC files) is being provided for the
1 km profiles. These are calculated using a refraction model and Satellite Toolkit (STK). If the
model calculation failed, it is recommended to use the 30 km tangent point (calculated
geometrically) information in the file headers. The GLCs can be used with the ACE-Imager and
ACE-MAESTRO profiles (interpolating in altitude as necessary). The file format for the GLC
files is provided in Table 6.
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 10
Recommendations for data use and error reporting:
The ACE Science Operations phase started on February 21, 2004. ACE measurements taken in
late 2003 and early 2004 were done as part of the Satellite Commissioning phase. Since February
21, 2004, there have been instances where there were issues with the data and the occultations
from these periods should be avoided or used with caution.
The occultations with known issues are listed on the ACE Data Issues webpage
(https://databace.scisat.ca/data_issues.php). This list is updated as new issues are reported so
please check it frequently. You can download a csv version of this using the button at the bottom
of the page.
If you find any problems or issues with the ACE data please let the ACE Science Operations Centre
know by submitting a Data Issue Report via:
(https://databace.scisat.ca/data_issues_report_form.php).
Recommendations for quality checking and filtering:
It is strongly recommended to perform quality checks on the data before finalizing your scientific
analysis. The data are not filtered in any fashion, and there is a significant probability of
encountering bad data, most commonly from altitude gaps in the measurements that stem from
spectra lost during transmission from the satellite to the ground. While we could provide data
quality flags, the broad range of conditions (season, geographic location, phenomena such as the
polar vortex and Asian monsoon, etc.) probed by the ACE-FTS measurements can lead to a large
degree of variability in the full data set for a particular molecule. Applying a filter on the full data
set might then require relaxed constraints to avoid removing real, valid results, which could allow
poor quality data to pass through the filter and impact your analysis. Tailoring the filter to the data
set being analyzed represents optimal treatment for the data.
There are a number of statistical approaches one can use to identify outliers. The simplest
of these is to find the standard deviation of the data at a given altitude and remove all results more
than three times the standard deviation away from the average. If large outlier “spikes” exist in
the set of retrieved profiles, an alternate approach might be more warranted, such as using the
median absolute deviation (MAD) rather than standard deviation. It can be helpful to remove very
large positive and large negative values first, and the “robust” statistical methods odten work
better. When working with larger data sets that may include multiple distributions within the data
(e.g., working with the results for a particular molecule in different seasons), then you might
consider median average deviation (MeAD) (see P.E. Sheese et al. “Detecting physically
unrealistic outliers in ACE-FTS atmospheric measurements”, Atmos. Meas. Tech., 8, 741-750,
2015).
Ideally the filter would remain as simple as possible. For example, if you are comparing
the results for a particular molecule in two different seasons, it might be preferable to apply
separate filtering for the two different seasons rather than treating the two seasons simultaneously
with a more complicated approach. The ideal filtering approach depends on the molecule (some
are more variable than others), what subset of the data is being analyzed, and perhaps the nature
of the analysis (if you are assessing variability, for example, you would need to be very careful not
to throw out too much data lest you introduce a bias).
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 11
We also suggest having a visual assessment of the data in conjunction with application of
the filter. The simplest approach is to generate a “spaghetti” plot (i.e., plot all of the profiles
simultaneously on a single graph) of all the data included in the analysis. Figure 1 shows such a
plot for all of the retrieved SO2 profiles from sunsets in May 2012 (in blue) plus two occultations
from May 2015 (in orange). The occultations from 2015 were included to make the point that
although those data would be readily removed via filtering, they are not actually bad data. The
two occultations exhibit enhancements in SO2 in the aftermath of a volcanic eruption, while the
data from 2012 has SO2 at background levels. If one were to blindly filter without first observing
the data, one might summarily discard the data of greatest scientific value without realizing it.
Sometimes the most interesting science is in the outliers.
Figure 1: All SO2 results from sunsets in May 2012 (in blue) plus data from a pair of occultations
in May 2015 (in orange).
Let us say, however, that we are actually interested in looking at background SO2 and
choose to remove the results showing signatures of enhanced SO2 from volcanic plumes. From
the remaining data, a visual inspection of Figure 1 might suggest that the results below 10 km
exhibit much higher variability, so maybe we should focus on the altitude region above that for
the first cut at our analysis. Figure 2 shows the data excluding the two enhanced SO2 occultations
and the results below 10 km.
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 12
Figure 2: All retrieved SO2 data above 10 km from sunsets in May 2012.
Simply looking at all of the data plotted simultaneously can readily illuminate outliers in
the data set. In Figure 2, there are only three points (circled) that are likely to be removed by
applying a filtering. In this example, a simple visual inspection allowed us to identify outliers
without the need for all the effort of implementing a statistical filtering routine.
Note that many of the data points in Figure 2 are negative. A negative VMR is physically
meaningless, but their presence is a symptom of the fact that background SO2 is so low. No
constraint is applied in the retrieval to yield positive definite VMR, and so when the retrieved
VMR is close to zero, negative values will often be obtained. This problem should be solved
through averaging the results from multiple occultations. If a measurable signal exists, the average
should come out to be positive.
Under no circumstances should you filter out negative VMRs unless they are very negative,
for example if the magnitude is more than twice the reported uncertainty. Filtering out negative
data will impart a positive bias to the results, as can be seen in Figure 3, where the data in Figure
2 were averaged with and without the negative values included.
Simply looking at the data being analyzed can be a very powerful tool in the analysis, from
identifying problem areas (for example, below 10 km in Figure 1), to identifying outliers (the
enhanced SO2 curves in Figure 1 and the circled points in Figure 2), to spotting systematic features
in the profiles. Again, it is a highly recommended step to include in the analysis.
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 13
Figure 3: The average SO2 from Figure 2 with and without negative VMR values included in the
average.
A few words on ACE-FTS altitude resolution:
The input aperture of the ACE-FTS instrument is 1.25 mrad. This angular extent subtends
an altitude range of 3-4 km at the tangent point. People commonly refer to this quantity as the
altitude resolution of the instrument, but that is not entirely accurate.
The ACE-FTS has a circular aperture, which serves to “weight” the measurement toward
the center of the field of view compared to what one would observe with a rectangular aperture.
Figure 4 depicts the ACE-FTS input aperture. The width of the field of view (FOV) is largest at
the center (indicated by arrow “A”), which corresponds to the tangent height assigned to the
measurement. Assuming plane parallel solar rays entering the instrument, contributions to the
signal from the atmospheric regions above and below the tangent height are suppressed because
the width of the FOV decreases as you move away from the center, for example in the vicinity of
the arrow B in Figure 4.
One can calculate the width of the FOV to generate altitude weighting functions (i.e., the
FOV width, relative to the center width, as a function of altitude for the given measurement) for a
set of ACE-FTS measurement to see how much they overlap, which gives a sense of the actual
altitude resolution for the instrument. Figure 5 shows the calculated results for a set of
measurements at latitude 45ºN with an altitude spacing of 2 km. Even though the tangent height
separations are well below the 3 to 4 km value typically reported for the instrument’s altitude
resolution, there is a clear separation in the weighting functions in Figure 5.
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 14
If one were to conservatively define the altitude resolution according to the half width of
the weighting functions, it would be the order of 2.5 km. Defining it according to having distinct
separation of the weighting functions would put the altitude resolution at less than 2 km.
B
A
Figure 4: A representation of the input aperture of the ACE-FTS instrument. The arrow A
indicates the width of the field of view at the center, which corresponds to the measurement tangent
height. The arrow B indicates the width of the field of view for solar rays associated with an
altitude above the tangent height.
Figure 5: Weighting functions (width of the field of view, relative to aperture center, as a function
of altitude for a given measurement) for a set of ACE-FTS measurements with 2 km spacing,
calculated for latitude 45ºN.
The same calculations performed for measurements with an altitude spacing of 1 km are
shown in Figure 6. Note the zoomed y-axis relative to Figure 5. There is still a degree of separation
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 15
in the peaks, but with this altitude spacing we now have the weighting functions from multiple
measurements overlapping, rather than just 2, as was the case in Figure 5. This has implications
in the stability for retrievals. Tangent height separations around 1 km occur at low altitudes, where
refraction effects compress the measurement spacing. Fortunately, in this situation, we typically
oversample and have multiple measurements within a 1 km span, allowing us to detangle the
information in the measurements.
Figure 6: Weighting functions (width of the field of view, relative to aperture center, as a function
of altitude for a given measurement) for a set of ACE-FTS measurements with 1 km spacing,
calculated for latitude 45ºN.
We do not need to worry about effects below 1 km. Our forward model calculations are
performed on a 1 km altitude grid, and if there are multiple measurements within a layer on the 1
km grid, we retrieve a single VMR value corresponding to the center of the layer rather than
multiples values corresponding to each of the measurement tangent heights.
Note that the calculations in Figures 5 and 6 assume that the measurements are snapshots,
whereas in reality the measurements are collected over a time span of 2 seconds, during which
time the pointing changes as the satellite progresses in its orbit. This “smears” the measurement
in altitude, where the degree of smearing depends on the rate of change of tangent height for the
occultation.
The finite FOV of the ACE-FTS instrument impairs our ability to derive fine structure in
retrieved profiles below the FOV width because our measurements will always represent averages
across the FOV. The finite FOV also complicates matters for spectral features that vary rapidly
with altitude, such the N2 continuum or H2O in the troposphere, where the assumption of the signal
varying roughly linearly across the FOV could perhaps suffer in accuracy. However, our ability
to determine the peak altitude of a VMR profile, for example, or the tropopause altitude [see M.I.
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 16
Hegglin et al., Validation of ACE-FTS satellite data in the upper troposphere/lower stratosphere
(UTLS) using non-coincident measurements, Atmos. Chem. Phys., 8, 1483-1499, 2008] is
governed by the altitude sampling rather than by the FOV.
In summary, the altitude resolution of 3 to 4 km typically reported for the ACE-FTS
instrument is likely overestimated. A value of 2 to 2.5 km would be more representative. Derived
profiles from ACE-FTS measurements will typically be smoother than reality (as a consequence
of the finite field of view), but the fidelity of the retrieved bulk-average variations with altitude
should be limited only by the altitude sampling, permitting one to determine altitude information
(e.g., the locations of peaks or minima) to precisions as high as 1 km.
2. Readme Files
ACE-FTS version 4.1
June 26, 2020
Please read the readme files for previous ACE-FTS processing versions.
In version 4.0, tangent height determination below 18 km had problems in the presence of
significant aerosol contributions to the spectrum. For version 4.1, the N2 continuum analysis
employed in determining these tangent heights was significantly improved. Non-Voigt line
shapes were used for CH4 and N2O lines in the region, far wing contributions from the nearby
nu3 band of CO2 were included in the calculations, and a “baseline slope” term was used in the
analysis to account for wavenumber variations in aerosol extinction.
In the pressure/temperature retrieval, saturation effects were avoided in the set of CO2
microwindows employed for version 4.1, fixing significant problems that using saturated lines
had caused in version 4.0.
In version 4.0, lower quality spectra (usually associated with a jump in pointing in the
vicinity of a cloud) caused spikes in retrieved VMR profiles. A filter was applied to remove
such measurements from the analysis for version 4.1.
An issue in pressure/temperature retrievals that caused intermittent failures in version 4.0
was fixed. Version 4.1 should experiences fewer “failed occultations” (occultations lost due to
failures in the pressure/temperature retrieval) than previous processing versions.
In the vicinity of the N2O5 spectral feature, non-Voigt parameters were determined for
N2O lines and re-determined for CH4.
For CH4 and N2O, non-Voigt parameters are employed in some of the microwindows.
Results on the tangent grid are not provided in v4.1, only on the 1 km grid.
Chris Boone
cboone@scisat.ca
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 17
ACE-FTS version 4.0
March 18, 2019
Please read the readme files for previous ACE-FTS processing versions.
When generating the Level 1 data (atmospheric transmittances) for version 4.0
processing, the high sun and deep space measurements used for calibration are now filtered for
excessive noise (unlike previous processing versions), which should reduce the variability in the
retrieved profiles for a number of occultations. The high sun and deep space measurements are
now averaged over +/- 3 orbits for the same occultation type (sunrise or sunset) rather than just
taken from the given occultation. This will recover occultations that had no reference spectra
(i.e., where reference spectra were lost during data downlink from the satellite, but the
atmospheric measurements were transmitted successfully) and reduce the noise in the calculated
transmittances.
There should be more occultations available in version 4.0 than were in previous
processing versions, thanks to the processing change described above, along with a number of
other changes. Please let us know of any occultations that appeared in previous versions but are
missing in version 4.0.
The failure rates for a number of molecules, particularly those using cross sections in the
analysis, should be significantly reduced in version 4.0.
In previous processing versions, under certain conditions (when the beta angle was very
large), it was common to get unphysical oscillations in the retrieved volume mixing ratio (VMR)
profiles. Measurement tangent heights were used for the retrieval grid, and the problems
occurred when the measurements got too close together. Version 4.0 processing also uses the
measurement tangent heights for the retrieval grid unless the tangent height separation becomes
too small, in which case the retrieval grid switches to an altitude sampling coarser than the
tangent height grid. The version 4.0 retrieval grid uses a minimum altitude spacing of 2 km for
tangent heights above 15 km, and a minimum spacing of 1 km for tangent heights below 15 km.
This limitation on the retrieval grid suppresses unphysical oscillations that commonly occurred
above 15 km in previous processing versions when the tangent height spacing dropped below 2
km. The data in the “tangrid” files are still provided on the measurement grid, but the results for
occultations that use the sparser retrieval grid are interpolated onto the measurement grid with
cubic splines.
A new source of solar (F10.7) and geomagnetic (ap) data was implemented for version
4.0. The later years of version 3.6 processing employed nominal values for these quantities
because the previous sources used for these data were discontinued. These quantities are used in
the calculation of high altitude pressure and temperature profiles -- used as inputs to the
pressure/temperature (P/T) retrieval -- from the MSIS software. Using observed values for these
quantities rather than nominal values should improve the shapes of the pressure and temperature
curves above the highest analyzed measurement in the P/T retrieval (~120 km). This should
improve retrieval results at high altitudes.
A pre-retrieval is performed for high altitude CO2 (between 110 and 140 km) to minimize
errors from the shape of the CO2 VMR profile above the highest analyzed measurement in the
P/T retrieval. Note that in some cases, the retrieved CO2 profile actually increases with
increasing altitude, which coincides with occultations where CO2 spectral features were observed
above 160 km, in measurements normally used as reference high-sun spectra, where one assumes
no contribution from atmospheric absorption. For version 4.0, measurements earmarked as
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 18
reference high-sun spectra are filtered out of the calculation if they contain atmospheric CO2
spectral features.
Tangent heights below 18 km in version 4.0 are determined from the N2 continuum.
Please report any anomalous behavior for this altitude region as we continue to assess the
implications of this change on the retrievals. Because CO2 is no longer used to generate the
tangent heights below 18 km, we can now retrieve low altitude CO2 from ACE-FTS
measurements. A set of flags has been added to the reported CO2 results indicating the nature of
the result at the given altitude. A flag of 1 indicates that the reported CO2 is in the altitude
region where CO2 was fixed in the P/T analysis to an a priori value calculated from a model
(usually between 18 and ~60 km). A flag of 2 indicates that the reported CO2 is in an altitude
region where the VMR profile for the molecule was retrieved (via an empirical function to
ensure smoothness) in the P/T analysis. Data in this region (usually in the region of ~60 to 125
km) may be used for scientific analysis. A flag of 3 indicates the reported CO2 is in the altitude
range above the highest analyzed measurement in the CO2 VMR profile retrieval (~125 km). A
flag of 4 indicates that the reported CO2 is retrieved with tangent heights fixed to the values
determined from the N2 continuum (between the lowest measurement and 18 km). These data
may also be used for scientific analysis. A flag of 5 indicates that the altitude is below the
retrieved range, simply a filler because the reported VMR value is the flag -999.
This processing version contains a number of new molecules not found in previous
versions, including HFC-134a (a column was reserved in the version 3 output files, but it was
never retrieved in that version), peroxyacetyl nitrate (PAN), acetone, CHF3, and CH3CN. SO2 is
also included in this version, but note that current background levels of this molecule are not
readily measured by the ACE-FTS. The data product is intended primarily for instances of
enhanced SO2, as is found in volcanic plumes. Similarly, ClO is also included in this version but
may only be usable for occultations probing elevated concentrations of the molecule during
chlorine processing events in polar spring.
This processing version also contains two new isotopologues: NO2 and HNO3 with
isotope 15N.
Although every effort has been made to reduce unphysical variability in the retrieved
VMR profiles for this processing version, there are situations that will cause large errors that
cannot be avoided, in particular altitude gaps resulting from data lost during downlink from the
satellite. It is important to evaluate the data when performing a scientific analysis, filtering
outliers through some statistical method, using standard deviation or median absolute deviation
to minimize the contribution of bad data on the results. We recommend making “spaghetti plots”
of profiles being averaged, a simple method to visually assess the data and readily spot outliers,
altitude regions of high variability, or systematic features.
Results are not constrained to be positive. For VMRs close to zero, it is common to
retrieve negative values. While this is a nonphysical result, one should not simply discard all
negative values when taking averages. Ideally the averaged result should come out to be positive
(or zero if there is no signal at the given altitude). Removing the negative values from the
average will introduce a positive bias. As a general rule, we suggest discarding negative values
only if the magnitude is more than twice the reported uncertainty.
Chris Boone
cboone@uwaterloo.ca
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 19
ACE-FTS version 3.6
December 16, 2016
There are no changes to the microwindow sets or to the algorithms employed in the
processing compared to version 3.5. The only changes to the software were to accommodate
porting from UNIX to a Linux-based architecture. The different computing environment will
lead to slightly different results than what would have been obtained on the UNIX system,
particularly for situations of low signal (e.g., near the upper altitude limit of a retrieval), but
typically well within the error estimate. For situations with strong signals, the differences for
results obtained on the two operating systems are expected to be negligible.
Version 3.5 data extend from February 2004 through March 2013. Version 3.6 data
cover November 2012 onward. During the period of overlap (November 2012 to March 2013),
version 3.6 results are provided only for occultations missed during the original version 3.5
processing of the time period, with no duplicated results between the two data sets. Version 3.5
and 3.6 are compatible data sets and may be combined in scientific analyses.
In version 3.5, the C2H6 results were generated by replacing the version 3.0 results with
the output from a C2H6 research product. The research product was incomplete, and not all C2H6
version 3.0 results were replaced. If a version 3.5 C2H6 occultation contains data above 20 km,
do not use the upper altitude data, although the data at lower altitudes can still be employed with
some caution. Version 3.6 is not affected by this issue.
Chris Boone
cboone@uwaterloo.ca
January 16, 2017
Recommendations for Upper Altitude Data Usage
Pressure and temperature profiles calculated from the US Naval Research Laboratory's
MSIS model, used in the ACE-FTS pressure/temperature retrieval, require F10.7 (a measure of
solar flux at wavelength 10.7 cm) and ap index (a measure of geomagnetic activity) as inputs. In
version 3.6 processing, measured ap index values were used only up to January 2013, and
measured F10.7 was used up to January 2015. Average values for the quantities were used
whenever current values were not available (i.e., after January 2013 for the ap index and after
January 2015 for F10.7). Errors in one or both of these parameters will impact retrieval results at
very high altitudes, above about 90 km. Expect systematic errors the order of 1 percent when
using average values for the ap index and a few percent when using average values for both
F10.7 and ap index. As such, it is not recommended to use ACE data for trend studies at very
high altitudes for time frames extending beyond January 2013.
Chris Boone
cboone@uwaterloo.ca
ACE-FTS version 3.5
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 20
August 26, 2014
From October 2010 onward, bad low-altitude pressure and temperature inputs had an adverse
affect on version 3.0 ACE-FTS results. Version 3.0 data from that time period should not be
used.
ACE-FTS version 3.5 fixes this problem by using better pressure and temperature inputs,
which were generated from the analysis run of the Canadian Meteorological Center’s “global
model.”
Research products from ACE-FTS version 3.0 were also incorporated into the version 3.5
results:
1) In version 3.0, the shape of the CO volume mixing ratio (VMR) profile above the highest
analyzed measurement was wrong, increasing too rapidly with increasing altitude, causing
problems near the upper altitude limit of the CO retrievals. In version 3.5, a constant VMR as a
function of altitude is assumed above the highest analyzed measurement.
2) In version 3.0, the microwindow for C2H6 extended up to 20 km, but there were additional
windows employed in the retrieval (intended to provide extra information for a number of
interferers in the C2H6 window and therefore improve the convergence stability) that extended up
to 22 km. The software therefore attempted to retrieve C2H6 up to 22 km. With minimal spectral
information between 20 and 22 km, the results in that altitude region were often wildly
oscillatory. For version 3.5, the upper altitude limits for the extra microwindows were lowered
to 20 km, removing the bad results between 20 and 22 km.
3) In version 3.0, HCFC-22 retrievals employed microwindows from two spectral regions: near
815 and 1115 cm-1. The region near 1115 cm-1 has a much higher signal-to-noise ratio, and
therefore this window dominated the retrieval, yielding an unexpected slope for the HCFC-22
results in the troposphere, likely a result of an unidentified interferer in the 1115 cm-1
microwindows. In version 3.5, the HCFC-22 retrievals only make use of the microwindow near
815 cm-1.
4) For some occultations in version 3.0 with low levels of tropospheric H2O, particularly in the
polar regions, the first guess for one of the interferers (H2O isotopologue 3, 17OH16O) was too far
off. As a result, the retrieval would “get stuck,” stopping without moving very far from the first
guess parameters. The first guess VMR profile for N2O is quite different from what you would
expect in the polar region. However, the N2O retrieval works fine for most (but not all
occultations) when the H2O retrieval is carried out first for a particular occultation, giving a
much better first guess for the weak H2O isotopologue 3 interferer. For version 3.5, N2O
retrievals were run a second time for all version 3.0 occultations, using the same software with
the same N2O microwindow set. For occultations after September 2010, H2O retrievals are
always implemented prior to N2O retrievals to ensure a good first guess for the weak interferer.
For occultations prior to October 2010, version 3.5 results will be the same as version 3.0
results except for the above four molecules. For occultations from October 2010 onward, the
version 3.0 results have been discarded, and the retrievals have been redone from scratch with
improved pressure and temperature inputs. The data from October 2010 onward employ the
exact same retrieval software as was used in version 3.0, except for CO, C2H6, and HCFC-22.
For version 3.5, the microwindow sets for C2H6 and HCFC-22 have been adjusted as described
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 21
above, and the shape of the CO VMR profile for altitudes above ~120 km has been changed as
described.
Chris Boone
cboone@uwaterloo.ca
ACE-FTS version 3.0
March 18, 2010
Please read the readme files for previous version of ACE-FTS processing.
In version 3.0 pressure/temperature retrievals, there should be a significant reduction in the
occurrence of unphysical oscillations in the retrieved temperature profiles compared to version
2.2. Also, version 3.0 removed the empirical function employed for pressure at low altitudes in
the pressure/temperature retrievals, which should remove a minor glitch near 23 km that was
sometimes observed in version 2.2 results.
The microwindow sets for all of the molecules have been upgraded. The maximum number of
interferers allowed in the software for version 3.0 is 17 (compared to 5 in version 2.2). When
multiple isotopologues of a given molecule serve as interferers in the retrieval, they are now
treated separately in version 3.0, with a different VMR profile attributed to each isotopologue.
Upper altitude limits for volume mixing ratio profile retrievals have been increased for most
molecules, pushing the retrievals as much as possible into the noise limit.
A CO2 volume mixing ratio (VMR) profile is retrieved in version 3.0. The differences between
this retrieved profile and the CO2 VMR profile derived during pressure/temperature retrievals
(employing the same set of CO2 microwindows) is used to assess limitations of the retrievals.
The error reported for the VMR retrievals on the 1 km grid is now a root-mean-square
combination of the least-squares (random) fitting error and the percentage difference observed
for CO2 in the given layer (i.e., the difference between the retrieved CO2 VMR and the value
employed in the pressure/temperature retrievals). Errors for data reported on the tangent grid are
still the straight least-squares fitting errors.
Version 3.0 VMR retrievals include the following molecules not available in version 2.2: COCl2,
COClF, H2CO, CH3OH, and HCFC-141b. The molecules HOCl and ClO were removed from
the list of molecules studied on version 2.2 due to a lack of quality. Columns that were always
empty in the version 2.2 outputs (CFC-113, HCFC-142b, etc.) will contain results in version 3.0.
The only column in the version 3.0 outputs that will remain empty (for the set of main
isotopologues) will be HFC-134a while I work on issues with the retrievals for that molecule. A
research product will follow for the molecule in the future.
There are results for many isotopologues in version 3.0, including subsidiary isotopologues of
H2O, CO2, O3, N2O, CO, CH4, and OCS.
Chris Boone
cboone@uwaterloo.ca
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 22
ACE version 2.2 O3 update
October 26, 2005
Preliminary validation efforts with ozone suggested that the ACE-FTS retrieval results
showed a low bias. The ozone microwindow set consisted of a set in the 1000-1150 cm-1 range
and a set in 1830-2130 cm-1 range. Upon closer inspection, it seems that the spectroscopic
information in the two regions is not entirely consistent. The latter microwindow set received a
higher weighting in the fitting process (because the SNR was higher in that region) and ended up
dominating the fit.
A new set of ozone microwindows was selected, restricting the selection to the 980-1130 cm-1
region. The software was upgraded to allow subsidiary isotopes as interferers. Ozone
isotopologues 2 and 3 were included as interfers for the updated ozone retrievals.
Tropospheric ozone results showed higher than expected variability. A method used to
accelerate the retrieval process runs into trouble where there are significant baseline effects.
The speedup was removed for the ozone update.
Chris Boone,
cboone@uwaterloo.ca
ACE version 2.2
May 24th, 2005
Please read the readme files from versions 1.0, 2.0, and 2.1 ACE-FTS processing.
The high altitude portion (i.e., above ~90 km) should be improved in this version.
The bug for the output on the retrieval grid (i.e., the tangrid files) has been fixed. One can use
either the results on the retrieval grid or the results on the 1-km grid.
The following weak molecules have been added to the processing: HOCl, H2O2, and HO2NO2.
This is a testing phase for these molecules. As with ClO, averaging results from different
occultations may be required.
The retrieval of subsidiary isotopologues begins with this version. Note, however, that there
appears to be a problem with HDO retrievals.
A change in the VMR retrieval approach made VMR profiles more susceptible to unphysical
oscillations in version 2.0. Care should be taken when comparing to the results for a single ACE
occultation. However, comparisons that employ average results from several ACE occultations
should not be strongly affected. This problem is not present for any other version of the ACEFTS processing.
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 23
August 24, 2005
Problems reported for version 2.2
In occultations with elevated levels of C2H6, there was on occasion a failure of the crosscorrelation approach used to align the calculated and measured spectra. With the given first
guess for the C2H6 profile, the measured and calculated spectra did not look similar enough for
the cross-correlation approach to work properly. This will be fixed in the next processing
version by increasing the microwindow width to include lines from interfering molecules, to
better constrain the cross correlation approach.
C2H2 was retrieved only for a small number of occultations. The software occasionally
crashed during C2H2 retrievals, and so it was taken out of the retrieval list.
Low altitude O3 (below ~10 km) sometimes shows variability higher than expected. An
approach used to speed up the processing reduced the effectiveness of the retrieval for molecules
with little information content at low altitudes when there were large baseline effects (i.e., the
baseline was not close to 1 and/or had a large slope). There could be problems for other
molecules with low information content at low altitudes such as HNO3 or HCl (i.e., molecules
with much higher VMRs in the stratosphere than in the troposphere), although this has not been
investigated fully.
Some occultations exhibited errors in temperature at high altitudes (above ~90 km). The
cause was compensating errors in the retrieved temperature and CO2 VMR. Be prepared to
discard some occultations when working above 90 km.
Some polar winter exhibited (likely unphysical) oscillations in the retrieved pressure and
temperature in the stratosphere. Note, however, that the errors should compensate and should not
translate to large errors in retrieved VMRs.
In the VMR retrievals, if there are two values reported in the lowest layer in the results on the
retrieval grid (the tangrid files), it is an output error. There should only be one retrieved quantity
in the middle of the layer. If, for example, it reports VMR values at 9.2 and 9.8 km, the value
reported at 9.2 km should be ignored, and the value reported for 9.8 km actually corresponds to
the middle of the layer (9.5 km).
Chris Boone
cboone@uwaterloo.ca
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 24
ACE version 2.1
May 24th, 2005
Please read the readme files from versions 1.0 and 2.0 ACE-FTS processing.
Version 2.1 processing was only performed on a subset of the measured occultations, mostly
concentrating on the Arctic measurements during January-March 2005. There was significant ice
contamination on the detectors during this time period. Results for some molecules are expected
to be noisier than usual, particularly HCN. ClONO2 below 18 km could also exhibit increased
noise.
The results on the retrieval grid (i.e., the "tangrid" files) did not always output properly. Use the
results on the 1-km grid. (Only these results were submitted to the AVDC). Note that the same
issue exists for version 2.0.
ClO was added to the retrievals. This is a very weak absorber, and so it may be better to average
results for several occultations with similar conditions rather than considering the results from a
single occultation. There only appears to be significant ClO present during the
Arctic spring occultations in this data set.
C2H2 does not appear to be processing properly.
Chris Boone
cboone@uwaterloo.ca
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 25
ACE version 2.0
January 20, 2005
Please read the ACE_readme.txt file from ACE-FTS version 1.0 processing. The setup of the
output files is the same as for version 1.0, although there are more molecules. Recall the papers
available for background information:
Bernath, P.F et al., Atmospheric Chemistry Experiment (ACE): mission overview, Geophys.
Res. Lett., submitted (2005)
Boone, C.D. et al., Retrievals for the Atmospheric Chemistry Experiment Fourier Transform
Spectrometer, Geophys. Res. Lett., submitted (2005)
Pre-prints of the papers can be found on the following Web site:
http://www.ace.uwaterloo.ca/data
In the T_fit column, 1 and 0 are used to replace T and F, respectively, from the version 1.0
output format.
Version 2.0 output files give results on both the standard 1-km grid and on the measurement
grid.
For version 2.0, problems encountered when measurement spacings were less than 1 km (the
altitude grid spacing) have been addressed.
A slightly improved approach is used for interpolating onto the 1-km grid for forward model
calculations. In version 1.0, you could get a (maximum 0.5 km) extrapolation that would serve
to slightly enhance unphysical oscillations in the results (when they were present).
For pressure/temperature retrievals below 25 km, an empirical expression with four
parameters is used for pressure retrievals (instead of using a parameter for each measurement).
For P/T processing, a bug was fixed whereby during retrievals below the "crossover", P and T
were fixed to the results of the retrieval above the crossover (rather than being fixed to the a
priori P and T).
The software was converted to use exclusively HITRAN molecule numbering (rather than
using ATMOS molecule numbering with the HITRAN 2004 linelist). A mismatch between the
assumed molecule numbering and the molecule numbers in the linelist caused some issues in the
troposphere (because of "phantom interferences").
The ability to retrieve subsidiary isotopologues was implemented in the software. As of
January 20th, 2005, the isotopologues were not being retrieved, awaiting completion of
microwindow selection. A second pass with the software will fill in the isotopologue results.
Note that HDO, which was included in the regular output files for version 1.0, will now be in a
separate file with all of the other subsidiary isotopologues.
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 26
Columns in the output files exist for some weak absorbers (HO2NO2, H2O2, HOCl, H2CO,
and HCOOH) that are not being retrieved. They will also be retrieved on a second pass of
processing, once I am comfortable with the ability to retrieve them reliably.
With a broader sample of atmospheric conditions available for evaluating microwindows, the
microwindow selection was revised to avoid instances of saturation. More microwindows were
added at low altitudes for several molecules to improve tropospheric results.
H2O: Microwindows changed to (1) avoid saturation experienced for some occultations, (2)
avoid the 3200 cm-1 region (which was strongly impacted by detector contamination), (3)
improve tropospheric retrievals, and (4) have fewer interferences in the multiple molecule
retrievals
O3: The upper altitude limit of the retrieval range was increased to 95 km. Microwindow
selection was redone to avoid significant interference from the 668 and 686 isotopologues
and to get more microwindows in the troposphere. More windows were also added in the
vicinity of the O3 concentration peak.
N2O: More microwindows at lower altitudes, particularly for the troposphere.
CO: Microwindows were adjusted improve results at low altitudes, particularly for the
troposphere.
NO2: The upper altitude limit was increased, mostly to capture the enhanced high altitude NOx
observed during February, 2004.
HCl: More microwindows were added, particularly at high altitudes.
COF2: Microwindows were adjusted to avoid residual solar features.
More lines were included in the retrieval.
SF6: The upper altitude limit was lowered to improve retrievals.
The following molecules have been added for version 2.0 that were not retrieved in version 1.0:
OCS, HCN, CF4, CH3Cl, C2H2, C2H6, and N2
Chris Boone
cboone@uwaterloo.ca
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 27
ACE version 1.0
September 11, 2004
Some issues to be aware of with the ACE data
In the vmr results, an entry of -999 indicates that no retrieval was performed at that altitude.
At high altitudes, above the highest measurement used in the analysis for a given molecule,
I include a VERY rough estimate of the molecule's vmr (it is a constant times the a priori value,
with the same constant used for all altitudes above the highest analysed measurement). These
data are flagged by the uncertainties being set to -888. Do not trust these results too far above
the highest analyzed measurement.
Pressure and temperature values were retrieved down to no lower than 12 km (the columun
labelled T_Fit indicates whether temperature was retrieved at that altitude: T for True and F for
False). Below 12 km, temperature and pressure were fixed to data from the Canadian
Meteorological Center.
High altitude results (above about 95 km) should be viewed with skepticism. The
temperature profiles above this altitude require further work.
No provision was made for identifying occultations with significant ice contamination on the
FTS detectors. Therefore, some occultations (particularly earlier ones) could experience a
deterioration of results at low altitudes, some molecules worse than others.
Uncertainties provided for the vmr results are statistical errors from the fitting process (1sigma), and do not include systematic contributions. A more detailed error budget will be
determined later.
The molecule NO sometimes has extremely low absorption through the mesosphere (increasing
for both higher and lower altitudes). For such occultations, the retrieved NO profile through the
mesosphere will look quite ugly. The results are to be ignored when this happens.
For occultations that cut out above 10-17 km (due to clouds), the bottom-most measurement
often gives results that are clearly out (presumably from the clouds affecting the measurement
just before the suntracker loses lock). Simply ignore the bottom point if it looks inconsistent.
For molecules with significant interferences (e.g., N2O5 and SF6), the vmr for the highest
analyzed measurement is sometimes suspiciously high. I am investigating the cause of this. If
you see a sharp increase in the highest retrieved points, don't trust it.
Chris Boone
Data usage guide and file format description for ACE-FTS version 4.1 (ASCII)
Page 28