Data Description

Version as of 08:11, 18 Aug 2018

to this version.

Return to Version archive.

View current version

This page describes the GGG2014 version of the TCCON Dry Column-Averaged Mixing Ratios of CO2, CO, N2O, CH4, H2O, HDO and HF. An overview presentation highlighting the updates from GGG2012 can be viewed in the TCCON_Update_GGG2014 attachment.

Our data use policy can be found on the Data Use Policy page.

The data can be downloaded from the TCCON data server, hosted by CDIAC.  These will be mirrored by Pangaea in the near future. The data are provided in netCDF format, and contain the retrieved values, and ancillary data like surface pressure, temperature, averaging kernels and a priori profiles. A MATLAB script to read the netcdf output is provided at the bottom of this page.

For the GGG2009 data description, please see the Data Description GGG2009 page; for the GGG2012 data description, please see the Data Description GGG2012 page.

For up-to-date information, please sign up to the TCCON Users mailing list.

Common questions about the usage of the data not answered below may be in our FAQ page.



The Total Carbon Column Observing Network (TCCON) column-average Dry Mole Fractions (DMF) were acquired using Fourier Transform Spectrometers (FTS) located throughout the world (see our sites page for their exact locations). Direct solar spectra are measured in the near-infrared spectral region, and these spectra were used to retrieve column abundances of CO2, O2, CH4, N2O, H2O, HDO and HF.


Dry column-average Gas G DMF has been calculated according to

While oxgyen abundance is declining due to combustion of fossil fuels, to within the precision of this equation, we take it to be fixed.

Where there are several bands from which a gas is retrieved, an error-weighted average is used for the final XG value. The standard windows used by TCCON retrievals are in the table in the Spectroscopy section below.

Detailed information about how to use our auxiliary data (averaging kernels and a priori profiles) can be found on the Auxiliary Data page.

Data Format

The data provided in the GGG2014 version will be in netCDF format. We have DOIs assigned to each dataset, and the DOI is embedded in the netCDF file, in the global attributes. For example:

Data_DOI = "XX.XXXX/tccon.ggg2014.jpl01.R0/YYYYYYY" ;

We ask that you cite the DOI for each TCCON dataset that you use in your publications.

The fields indexed by time include the retrieved values: xco2_ppm and its error, xch4_ppm and its error, xn2o_ppb and its error, xco_ppb and its error, xhf_ppt and its error, xh2o_ppm and its error, xhdo_ppm and its error. There are also ancillary data indexed to time: latitude and longitude (lat_deg, long_deg), fvsi (the fractional variation in solar intensity, which is a proxy for cloudiness), solar zenith and azimuth angles (asza_deg, azim_deg), several meteorological fields (wind speed and direction, surface pressure, temperature and humidity), and the GSETUP and GFIT versions.

We have included the a priori profiles in the netCDF files. The a priori profiles are generated once per day, and so there are far fewer a priori profiles than retrieved values. To assist you in matching up a retrieved value to its appropriate a priori profile, we have included a field called prior_date_index that has the dimension of time to map between the a priori profile and the retrieved value. Important: please note that the index starts at 0!

The Lamont column averaging kernels are also included in the public files, and they are binned into 5-degree solar zenith angle bins on 71 pressure levels. Please interpolate the column averaging kernel to the solar zenith angle of the measurement(s) you wish to compare using the asza_deg field.

More information about how to use our auxiliary data can be found on the Auxiliary Data page.

An example header for a TCCON netCDF file can be found on the netCDF File Header Example page.

Back to top

Sources of Uncertainty

The dry column-averaged DMF is affected by three main sources of uncertainty:

Measurement precision (repeatability)

The TCCON measurement precision varies from site to site by is generally <0.25% (1-sigma) for single measurements (approx 90 s) XCO2 under clear or partly cloudy skies (up to 5% fractional variation in solar intensity), and solar zenith angles <82 degrees. For XCH4, it is <0.3%; XCO is <1%; XN2O is <0.5%.

Spectroscopic uncertainties

The effects of spectroscopic uncertainties are, to first order, removed by our airmass correction and in situ calibration. There are second-order effects that undoubtedly remain. The absolute accuracy of the retrievals was calibrated by comparison to integrated aircraft and AirCore profiles, resulting in the corrections listed below in the corrections and calibrations section. The aircraft profiles are generally performed at lower airmasses (low SZA), and the column-average DMF are now well-calibrated for these values. However, additional systematic uncertainties may be present at higher airmass due to uncertainties in the spectroscopy of the gas of interest and/or O2.

Systematic instrumental changes over time

There are known drifts and changes in these instruments, due to the ILS and ghosts. Care is taken to minimize and correct for these uncertainties. As discussed below, GGG2014 explicitely reduces the influences of ghosts.  The next version of GGG will account for changes in the ILS due to zero level offsets and imperfect alignment. 

Error Budget

The following is a rough error budget for the GGG2014 data and a comparison to the GGG2012 data.
Error, Cause, and Test XCO2 XCH4 XCO XN2O XH2O XHDO XHF
GGG Version 2012 2014 2012 2014 2012 2014 2012 2014 2012 2014 2012 2014 2012 2014

Laser Sampling Errors ("ghosts")*


Laser mis-sampling

0-0.25% <0.025%  0-0.25%  <0.025%  0-0.25%  <0.025%  0-0.25%  <0.025%  0-0.25%  <0.025%  0-0.25%  <0.025%  0-0.25%  <0.025%

Zero Level Offsets*


Detector non-linearity


0.1% change in ZLO

0.03% 0.03%   0.05%   0.1%   0.08%   0.005%   0.07%   0.15%

ILS errors*


Optical misalignment

0.1% 0.1%                        

Smoothing Error


A priori profile shape


CO2: assume an invariant a priori profile shape with altitude

CH4, CO, N2O, HF: adjust the a priori profile up 1 km

H2O, HDO: halve the a priori value in the lowest layer of the atmosphere

0.06% 0.06% 0.05% 0.05% 1% 1% 0.3% 0.3% 0.6% 0.6% 0.9% 0.9% 2% 2%

In Situ Calibration


Aircraft ceiling altitude
Aircraft measurement error


CO2, CH4, N2O, CO: Shift up the stratospheric data by 1 km above the aircraft/AirCore ceiling; add in quadrature to aircraft measurement error

H2O: Assume 5% error on radiosonde measurements

Np is the number of profiles used in the comparison

0.1% (Np=16) 0.07% (Np=32) 0.45% (Np=9) 1.04% (Np=26) 7.25% (Np=9) 7.19% (Np=21) 1.2% (Np=7) 1.85% (Np=8) 5% (Np=30) 5% (Np=30)        

Atmospheric a priori temperature


NCEP, time interpolation


Add 1 K to temperature profile

0.04% 0.04%   0.05%   0.15%   0.1%   0.3%   0.1%   0.18%

Atmospheric a priori pressure


NCEP, time interpolation


Subtract 0.1% from pressure profile
0.06% 0.06%   0.035%   0.1%   0.07%   0.04%   0.03%   0.06%

Surface Pressure


Faulty or inaccurate pressure sensor


Add 1 hPa to the surface pressure measurement

  0.04%   0.035%   0.12%   0.07%   0.04%   0.05%   0.02%

Random Noise


Mainly solar photon noise (N=number of spectra averaged)

0.25N-0.5% 0.25N-0.5% 0.3N-0.5% 0.3N-0.5% 1N-0.5% 1N-0.5% 0.5N-0.5% 0.5N-0.5%            

*Denotes error terms that produce site-dependent biases

Back to top

Data Filtering

TCCON data are filtered on several criteria. The strictest of these is the cloudiness during a measurement, represented by the "fvsi" (fractional variation in solar intensity) parameter, which is the change in solar intensity over the course of a measurement (standard deviation over ~90 seconds), divided by the average solar intensity during the measurement. The fvsi parameter is expressed in percentage, and we filter out data with fvsi>5%. Other filters are much less strict, and exclude data with very poor spectral fits. Finally, individual sites filter data that are known to be suspect, and these are described in the site-specific notes below.

Data Comparisons

TCCON data should be compared with other data using the Rodgers and Connor (2003) approach. Column averaging kernels and a priori profiles are available as part of the netCDF files for the GGG2014 version of the data. Each file has a set of column averaging kernels interpolated into 5-degree solar zenith angle bins from a set of Lamont averaging kernels. The averaging kernels from other sites are negligibly different. A detailed description of how to use the averaging kernels and a priori profiles are on the Auxiliary Data page. Individual principal investigators can also help you out.

Changes from the GGG2012 Version

The differences between GGG2012 and GGG2014 are summarized in the GGG2014 TCCON Update presentation attached at the bottom of this page. Briefly, the following changes have occurred between GGG2012 and GGG2014:

Laser Sampling Errors

  • An algorithm for correcting laser sampling errors in older TCCON data has been implemented and no further corrections to the data should be applied to account for these errors.

  • At some sites, the laser sampling errors changed over time producing a bias in Xgas that also changed over time; at other sites, changes in the laser sampling error were typically step changes that occurred when lasers were changed or electronics boards were switched out.

  • Unaccounted for laser sampling errors likely yield residual biases that are less than 0.025%.


  • We have made many changes to the H2O spectroscopy in the 4000-6000 cm-1 region based on Kitt Peak cell spectra. This affects retrievals in the H2O, HDO, CO and N2O windows.
  • A 13CH4 line list from HITRAN2012 (plus small modifications) was added. This impacts retrievals in the CO and CH4 windows.
  • The HITRAN2012 line list for CO was adopted. Combined with the H2O and 13CH4 spectroscopic changes, the CO retrievals are much improved for GGG2014.
  • An additional retrieval window (centered at 4719.50 cm-1) was added for N2O.
Retrieval Spectral Windows

Where there are several bands from which a gas is retrieved, an error-weighted average is used for the final XG value. The standard windows used by TCCON retrievals are in the table below. New windows added for the GGG2014 version are in orange. Warning: Some TCCON stations can not measure all of the windows (e.g., CH4, H2O, HDO for Tsukuba120, H2O, HDO for Karlsruhe and Izana, see the site-specific comments for details). Since there are small systematic biases between windows of the same gas, results from sites that do not use the full set of windows may be biased relative to the broader network data. This DOES NOT influence CO2 measurements.

Gas Central Wavenumber
Window Width
  Gas Central Wavenumber
Window Width
CO2 6220.00
  CO 4233.00
CH4 5938.00
  HF 4038.95 0.32
N2O 4395.20
  O2 7885.00 240.00
H2O 4565.20
  HDO 4054.60


Bias Correction

"In situ" (airmass independent) and airmass dependent bias correction factors were applied as listed in the CORRECTIONS AND CALIBRATIONS section below. The GGG2009 values, for comparison, can be found on the GGG2009 CORRECTIONS_AND_CALIBRATIONS page; the GGG2012 values can be found on the GGG2012 CORRECTIONS AND CALIBRATIONS page.

  • In the GGG2014 calculation of the airmass independent bias correction factors, we added many more aircraft and aircore profiles to the calibration curves. This, together with the correction of the laser sampling errors, slightly changed the slope of the calibration curves. In addition, to improve precision we increased the number of significant figures used in describing the fit of the calibration curve. Together, these resulted in a global change in XCO2 of -0.3 ppm.

Overall Differences from GGG2012

  • As a result of the corrections listed above, the retrieved values of Xgas have changed relative to the GGG2012 analyses:
    • XCO2 decreased by 0.3 ppm (0.075%)
    • XCH4 decreased by ~1.5-2ppb (0.08-0.1%)
    • XCO changes differ from site-to-site due to the spectroscopic and a priori profile improvements - there is a seasonality to the differences as well
    • XN2O is now generally higher by <1 ppb (<0.3%) with a small seasonal cycle that peaks in the summer
    • XH2O is now generally higher (<150 ppm; ~1%) with a seasonal cycle that peaks in the summer
    • XHDO is now generally slightly higher (<15 ppm; ~0.2%) with a seasonal cycle in the northern hemisphere
    • XHF has a small overall difference (<0.5 ppt; <0.8%) that differs from site to site

Back to top


The data are corrected for known airmass-dependent and airmass-independent (calibration) biases. These correction factors are applied to the column-averaged mole fractions.

The airmass-dependent correction factor (ADCF) is determined offline from the symmetric component of the diurnal variation. The airmass-independent correction factor (AICF) is determined offline by comparisons with in situ profiles measured over TCCON sites from aircraft or balloon payloads.

Gas ADCF ADCF Error (σ) AICF AICF Error (σ)
xco2 -0.0068 0.0050 0.9898 0.0010
xch4  0.0053 0.0080 0.9765 0.0020
xn2o  0.0039 0.0100 0.9638 0.0100
xco -0.0483 0.1000 1.0672 0.0200
xh2o - - 1.0183 0.0100

No airmass dependent correction is applied to XH2O, XHDO or XHF.

The airmass-dependent calibration is described in the supplementary material to:

Wunch, D., G. C. Toon, J.-F. L. Blavier, R. A. Washenfelder, J. Notholt, B. J. Connor, D. W. T. Griffith, V. Sherlock, P. O. Wennberg. The Total Carbon Column Observing Network, doi: 10.1098/rsta.2010.0240 Phil. Trans. R. Soc. A 28 May 2011 vol. 369 no. 1943 2087-2112. Available from: http://rsta.royalsocietypublishing.o.../2087.full.pdf

The method for determining the in situ calibration factors was developed and described in:

Wunch, D., Toon, G. C., Wennberg, P. O., Wofsy, S. C., Stephens, B. B., Fischer, M. L., Uchino, O., Abshire, J. B., Bernath, P., Biraud, S. C., Blavier, J.-F. L., Boone, C., Bowman, K. P., Browell, E. V., Campos, T., Connor, B. J., Daube, B. C., Deutscher, N. M., Diao, M., Elkins, J. W., Gerbig, C., Gottlieb, E., Griffith, D. W. T., Hurst, D. F., Jimenez, R., Keppel-Aleks, G., Kort, E. A., Macatangay, R., Machida, T., Matsueda, H., Moore, F., Morino, I., Park, S., Robinson, J., Roehl, C. M., Sawa, Y., Sherlock, V., Sweeney, C., Tanaka, T., and Zondlo, M. A.: Calibration of the Total Carbon Column Observing Network using aircraft profile data, Atmos. Meas. Tech., 3, 1351-1362, doi:10.5194/amt-3-1351-2010, 2010. Available from:

Messerschmidt, J., M. C. Geibel, T. Blumenstock, H. Chen, N. M. Deutscher, A. Engel, D. G. Feist, C. Gerbig, M. Gisi, F. Hase, K. Katrynski, O. Kolle, J. V. Lavric, J. Notholt, M. Palm, M. Ramonet, M. Rettinger, M. Schmidt, R. Sussmann, G. C. Toon, F. Truong, T. Warneke, P. O. Wennberg, D. Wunch, and I. Xueref-Remy (2011), Calibration of TCCON column-averaged CO2: the first aircraft campaign over European TCCON sites, Atmospheric Chemistry and Physics, 11(21), 10765-10777, doi:10.5194/acp-11-10765-2011. Available from:

Geibel, M., J. Messerschmidt, C. Gerbig, T. Blumenstock, H. Chen, F. Hase, O. Kolle, J. V. Lavric, J. Notholt, M. Palm, M. Rettinger, M. Schmidt, R. Sussmann, T. Warneke, and D. G. Feist (2012), Calibration of column-averaged CH4 over European TCCON FTS sites with airborne in-situ measurements, Atmos. Chem. Phys., 12, 8763–8775, doi:10.5194/acp-12-8763-2012. Available from:

Note that XHF and XHDO are not calibrated. These data sets should be used with caution in any analysis that requires absolute measurements.


TCCON retrievals of XHF are uncalibrated because there have been no WMO-standard profiles of HF measured over our TCCON stations. The XHF asbsorption line used for TCCON retrievals is narrow and on the wing of a strong H2O line. This line can black out when it's wet, especially at high airmass. If you wish to use these data, please ensure that you carefully inspect the data and use low airmasses only. It would be appropriate for you to consult with a TCCON PI about the use of the XHF data. These XHF retrievals are useful if you wish to compute tropospheric columns of CH4, and perhaps N2O:

Washenfelder, R. A., P. O. Wennberg, and G. C. Toon (2003), Tropospheric methane retrieved from ground-based near-IR solar absorption spectra, Geophysical Research Letters, 30(23), 1-5, doi:10.1029/2003GL017969. Available from:

Saad, K. M., Wunch, D., Toon, G. C., Bernath, P., Boone, C., Connor, B., Deutscher, N. M., Griffith, D. W. T., Kivi, R., Notholt, J., Roehl, C., Schneider, M., Sherlock, V., and Wennberg, P. O.: Derivation of tropospheric methane from TCCON CH4 and HF total column observations, Atmos. Meas. Tech. Discuss., 7, 3471-3501, doi:10.5194/amtd-7-3471-2014, 2014. Available from: http://www.atmos-meas-tech-discuss.n...3471-2014.html.

Wang, Z., Deutscher, N. M., Warneke, T., Notholt, J., Dils, B., Griffith, D. W. T., Schmidt, M., Ramonet, M., and Gerbig, C.: Retrieval of tropospheric column-averaged CH4 mole fraction by solar absorption FTIR-spectrometry using N2O as a proxy, Atmos. Meas. Tech. Discuss., 7, 1457-1493, doi:10.5194/amtd-7-1457-2014, 2014. Available from: http://www.atmos-meas-tech-discuss.n...1457-2014.html


TCCON retrievals of XHDO are uncalibrated because there have been no WMO-standard profiles of HDO measured over our TCCON stations. These data, however, have been used in several papers in which the absolute HDO value is not critical. We would recommend that you use caution when using these data. Below are several references that have used previous versions of TCCON HDO. See the HDO FAQ for details.

Risi, C., D. Noone, J. Worden, C. Frankenberg, G. Stiller, M. Kiefer, B. Funke, K. Walker, P. Bernath, M. Schneider, D. Wunch, V. Sherlock, N. Deutscher, D. Griffith, P. O. Wennberg, K. Strong, D. Smale, E. Mahieu, S. Barthlott, F. Hase, O. García, J. Notholt, T. Warneke, G. Toon, D. Sayres, S. Bony, J. Lee, D. Brown, R. Uemura, and C. Sturm (2012), Process-evaluation of tropospheric humidity simulated by general circulation models using water vapor isotopologues : 1 . Comparison between models and observations, Journal of Geophysical Research, 117, 1-26, doi:10.1029/2011JD016621. Available from:

Risi, C., D. Noone, J. Worden, C. Frankenberg, G. Stiller, M. Kiefer, B. Funke, K. Walker, P. Bernath, M. Schneider, S. Bony, J. Lee, D. Brown, and C. Sturm (2012), Process-evaluation of tropospheric humidity simulated by general circulation models using water vapor isotopic observations: 2 . Using isotopic diagnostics to understand the mid and upper tropospheric moist bias in the tropics and subtropics, Journal of Geophysical Research, 117(D05304), 1-25, doi:10.1029/2011JD016623. Available from:

Frankenberg, C., D. Wunch, G. Toon, C. Risi, R. Scheepmaker, J.-E. Lee, P. Wennberg, and J. Worden (2013), Water vapor isotopologue retrievals from high-resolution GOSAT shortwave infrared spectra, Atmospheric Measurement Techniques, 6(2), 263–274, doi:10.5194/amt-6-263-2013. Available from:

Boesch H., N. M. Deutscher, T. Warneke, K. Byckling, A. J. Cogan, D. W. T. Griffith, J. Notholt, R. J. Parker, and Z. Wang (2013), HDO/H2O ratio retrievals from GOSAT, Atmospheric Measurement Techniques, 6(3), 599-612, doi:10.5194/amt-6-599-2013. Available from

Back to top



The wind speed and direction sensor was faulty until March 26, 2009. It is possible that the wind speed values prior to March 26, 2009 are valid but noisy, but the wind direction values are not. All values since March 26, 2009 are valid.


This instrument was at JPL to provide simultaneously recorded atmospheric spectra during the original OCO thermal-vacuum testing period. The instrument was subsequently moved to Lamont, Oklahoma. The OCO flight instrument was lost due to launch failure.

The wind speed and direction sensor was faulty for the entire time series. It is possible that the wind speed values are valid but noisy, but the wind direction values are not.

Wunch, D., P. O. Wennberg, G. C. Toon, G. Keppel-Aleks, and Y. G. Yavin (2009), Emissions of greenhouse gases from a North American megacity, Geophysical Research Letters, 36(15), 1-5, doi:10.1029/2009GL039825. Available from:


This instrument was at JPL to provide simultaneously recorded atmospheric spectra during the OCO-2 thermal-vacuum testing period (see the Frankenberg et al. for OCO-2 TVac testing details). The instrument was subsequently moved to Indianpolis, Indiana, for the INFLUX campaign before returning to JPL for comparison with the OCO-2 instrument spare (OCO-3). It is currently located next to the dry lake bed at Edwards, CA (also referred to as NASA-Dryden and subsequently NASA-Armstrong).

Back to top

Park Falls

The first TCCON station was installed in Park Falls in May, 2004. The paper describing the station is:

Washenfelder, R. A., G. C. Toon, J.-F. L. Blavier, Z. Yang, N. T. Allen, P. O. Wennberg, S. A. Vay, D. M. Matross, and B. C. Daube (2006), Carbon dioxide column abundances at the Wisconsin Tall Tower site, Journal of Geophysical Research, 111(D22), 1-11, doi:10.1029/2006JD007154. Available from:

There are several data gaps due to failure of the solar tracker enclosure and of the Bruker IFS125 scanner. 

Back to top

Ascension Island

Solar tracker mirror degradation in 2012 and 2013

Ascension Island is a very rough environment. The solar tracker mirrors suffered from degradation due to sea salt corrosion and dust from the very beginning (May 22, 2012). The reflectivity of the mirrors was reduced from nearly 100% to about 30-40% in the infrared (25% in the visible). This reduced the number of spectra per day and also the signal-to-noise-ratio of the spectra that were measured. Still, about 50% of the measured spectra (the ones that were submitted to the TCCON database) appeared to be usable for further analysis.

The 2nd measurement period with new protected-gold mirrors started from mid-March 2013. The new mirrors performed much better than the first set at the beginning. Howver, they still suffered from the harsh environment. By the end of June 2013, the mirrors were too degraded for further measurements. The instrument was then shut down until the next planned maintenance in August 2013.

The problem was finally solved with the introduction of specially designed rugged mirrors in August 2013. There are three distinct periods with respect to the different types of mirrors that were used:

  • 20120522-20120831: original unprotected gold mirrors (no cleaning)
  • 20130317-20130618: protected gold mirrors (cleaning every 1-2 weeks)
  • 20130911-today:      rugged mirrors (cleaning every 1-2 weeks)

The mirrors need to be cleaned regularly to remove dust from the surface. Otherwise, the solar tracker would become unable to track the sun after 3-4 weeks. The effects of cleaning may be visible in the data, for example in the form of a sudden decrease in noise or an increase in the number of measurements per day.

Back to top

Reunion Island

The Reunion Island XCO2 data have a residual airmass dependence even after the known TCCON-wide airmass dependence is removed. The cause of this is still under investigation.

From August 2 2012 until November 12 2012, a linearly increasing time correction (tcorr) was applied to compensate for a deviating computer clock which was not receiving NTP time updates during this period.

The daily spread of the retrieved Xgas from November 26 2012 until December 13 2012 is considerably larger due to a degrading laser tube. To reflect this, the errors for Xco2 are about three times larger during this period compared with the errors obtained during optimal operation.
The bad laser tube was replaced on February 18 2013.

From October 10, 2013 until November 20, 2013 a 0.8 mm iris was used for the measurements (instead of 0.5 mm for the remaining data). This resulted in saturation of the detector at low solar zenith angles. The measurements which were affected by saturation have been excluded from the public data set based on the value of the solar intensity average (SIA).  [SIA is the mean detector voltage during a measurement and at Reunion has a range of 0 V to 0.7 V.  Measurements with an SIA > 0.5 V were excluded due to saturation].

Back to top


Some data before 5 July 2010 originally impacted by detector saturation have been excluded from this release. 

Back to top


The JPL2011 instrument was moved to Indianapolis, IN as part of the INFLUX campaign and recorded spectra from August 23, 2012 to December 1, 2012.  It subsequently returned to JPL for thermal-vacuum testing of the OCO-2 flight spare, and is now in Dryden/Edwards/Armstrong.

Back to top


To protect gold mirrors of the solar tracker from degradation, a high transmission glass is used for a cover of the solar tracker.

Back to top


The JPL2011 instrument is now located at NASA's Armstrong Flight Research Facility (Edwards Air Force Base) in support of OCO-2 and has been operational since July 20, 2013. This site is also often called "Dryden", the previous name for this NASA center. 

Back to top


The Pasadena/Caltech instrument has been running since September 20, 2012. This instrument has been through several lasers, most of which failed due to faulty manufacturing. As a result, the data between April 30 and May 22, 2013 are much noisier than the rest due to a slow laser failure.

A problematic motor that drives the scanning mirror in the FTS caused a sometimes significant difference in retrieved Xgas values between forward and reverse scans. The motor was replaced on April 20, 2015, and data after that date are unaffected by this problem. To mitigate the error in the 20120920-20150419 time series, we have released a new version of the data (R1) that minimizes these errors by averaging forward and reverse scans. Data recorded on and after April 20, 2015 have not been averaged.

An additional problem with the R0 version of the data was a poorly calibrated pressure sensor. The calibration error (~3.8hPa) causes a bias in XCO2 of 0.6 ppm and in XCH4 of 3 ppb, with the R1 data being lower than the R0 data. This problem has been resolved in the R1 version.

Back to top


The spectrometer has been running operationally since April, 2010. The setup pioneers what may become the TCCON-next-generation configuration, as it covers NIR and MIR with the ability of simultaneous measurements in the NIR and an additional MIR spectral range. The initial design did not cover the complete spectral range covered by the extended InGaAs - for this reason not all TCCON data products are currently provided. However, in August 2012 an optical filter and the InGaAs diode have been replaced and since this date the spectrometer observes the complete InGaAs spectral range. The next data release for Karlsruhe will include all TCCON data products from August 2012 onwards.

The particular optical setup of the Karlsruhe FTIR instrument induces stronger variations of the background continuum level in solar spectra which are not of atmospheric origin than the standard setup. Therefore, several spectral windows are not well fitted in the continuum level by the standard GGG2014 retrieval and residuals follow atypical shapes. To mitigate the error, we have released a new version of the Karlsruhe Data Set (R1) that minimizes these error by fitting curvatures in the continuum level using discrete Legendre polynomial basis functions, which is a feature available in GGG2014, but not typically needed for TCCON spectra. Data from April 19, 2010 to December 28, 2014 have been reprocessed.

Back to top


The Darwin instrument is a Caltech-built container and has been running since September 2005. Interruptions and changes to the operation are summarized below (for details, see the DarwinEventLog.txt attached below).

2005-09-05    Start measurements
2005-10-27    Start DC recording of interferograms
2005-12-01    Replace degraded tracker mirrors
2006-02-04    TWP-ICE calibration overflight
2006-07-12 to 2008-11-30
                    Various periods of unexplained O2 offsets after instrument resets
2009-01-12    Infrared realignment (non-fringes method)
2009-04-21    Laser unstable, period of reduced data density.
2009-09-18    Replaced laser, not fully collimated until 2012-10-22
2009-10-22    Tracker realignment
2010-01-04    Tracker failure, no data
2010-09-05    Tracker reinstalled
2010-10-25    Tracker realigned
2011-07-01 to -04    HIPPO IV overflights
2012-01-22    Replace laser board to ECL04, minimise ghosts
2012-05-13    Re-minimise ghosts
2012-10-22    Full IR and laser realignment, fringe method.
2014-01-20    Ghost minimisation

Back to top


Routine TCCON measurements started in June 2008. Additional meteorological parameters (wind speed/direction, outside temperature and humidity) were recorded starting in 2009 Feb 20. The dates below summarize the important events related to the operation, which could affect the data (for more details, see the WollongongEventLog.txt attached below).

2008-06-27    Start routine measurements
2009-01-07    Replace tracker mirrors
2009-03-16 to 2009-07-25
                     Beamsplitter misalignement, all data flagged.
2009-07-25    Realignment, IR and laser.
2010-04-15 to 2010-06-30
                     Misalignment, reason unclear. All data flagged.
2010-06-30    IR corner cube and laser realignment, fringe method
2011-07-22    Replace laser ECL04 boards, align laser, minimise ghosts
2012-10-02    Full realignment, fringe method
2013-05-15    Ghost minimisation
2014-08-03    Laser died
2014-08-19    Laser replaced (REO), restart measurement

Back to top


Routine TCCON measurements began at Lauder in June 2004 using a 120HR spectrometer. In 2009 a 125HR spectrometer was installed at the site and has been the dedicated TCCON instrument since February 2010. The initial GGG-2014 release of Lauder datasets covers the following periods:

  • 120HR: 2004-06-29 to 2010-12-09
  • 125HR: 2010-02-02 to 2014-06-30

The 125HR timeseries has been resampled using the standard TCCON procedures for the GGG-2014 release. The 120HR timeseries has not been resampled. There are small systematic differences between the two timeseries. Refer to attached release notes for further details.

Back to top


The TCCON site at Bialystok has been operational since March 2009. Through the first ~6 months, the data were affected by a large laser sampling error, which caused xCO2 errors of 0.5ppm, and was present in the GGG2012 release. The data have subsequent had this effect removed, and are ghost free.

Several data gaps exist in the time series. These relate to laser failure, and intermittent problems with the tracker cover.

Numerous timing errors in the data have been corrected examining the NTP logs and the symmetry of the xAIR retrievals as a function of solar zenith angle. The maximum xGAS error caused by the timing errors is < 0.025%.

With the R1 data, the time correction has been updated by taking the spectra start-time from stored log-files instead of the IFS start-time, which turned out to be correct in almost all cases. Only in 2009 and 2010 the NTP didn't work sometimes cause of bad internet, this has been corrected as before.

Back to top


Measurements have been taken at Orleans since late August 2009. Like Bialystok, data gaps exist due to laser failures, and occasional tracker cover problems.

Timing errors have been corrected as for Bialystok.

Back to top


•Tsukuba 125 HR (new FTS), two-letter site name: tk (Apr. 08, 2010 -present)

New FTS (IFS 125 HR) started operation from Apr. 08, 2010. However, spectra were affect by a laser sampling error from Apr. 08, 2010 to Jul. 28, 2011 and by a lot of aftershock after a very huge earthquake with 9.0 magnitudes occurred on Mar. 11, 2011 from Apr. 11, 2011 to Jul. 28, 2011. Therefore, we are releasing the data without ghost and earthquake effects from Aug. 4 2011. We expect it will tike more time to release data before Aug. 4 2011.

There are two types of archived data for the Tsukuba site. Please use the data of Data_Revision of R1.
Data_Revision of R0 was retrieved with GIFT of GGG2014 but the post-processing was performed with correction factors used in GGG2012. This is a post-processing mistake.

Data_Revision of R1 was retrieved with GIFT of GGG2014 and the post-processing was performed with correction factors used in GGG2014.

•Tsukuba 120 HR (old FTS, now at Rikubetsu after upgrading), two-letter site name: tj (Dec. 25, 2008 - Mar. 22, 2010)

In data analysis using GGG2014, it was found it was not easy to process Tsukuba 120 HR spectra (old OPUS file formant and AC sampling). We expect it will take considerable time to release data.


There are two types of archived data for the Rikubetsu site. Please use the data of Data_Revision of R1.
Data_Revision of R0 was retrieved with GIFT of GGG2014 but the post-processing was performed with correction factors used in GGG2012.This is a post-processing mistake.

Data_Revision of R1 was retrieved with GIFT of GGG2014 and the post-processing was performed with correction factors used in GGG2014. 

Back to top


This site has been viewed 25322 times.