Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

Atmospheric Concentrations of Greenhouse Gases

Identification

1. Indicator Description

This indicator describes how the levels of major greenhouse gases (GHGs) in the atmosphere have
changed over geological time and in recent years. Changes in atmospheric GHGs, in part caused by
human activities, affect the amount of energy held in the Earth-atmosphere system and thus affect the
Earth's climate. This indicator is highly relevant to climate change because greenhouse gases from
human activities are the primary driver of observed climate change since the mid-20th century (IPCC,
2013).

Components of this indicator include:

• Global atmospheric concentrations of carbon dioxide over time (Figure 1).

• Global atmospheric concentrations of methane over time (Figure 2).

• Global atmospheric concentrations of nitrous oxide over time (Figure 3).

• Global atmospheric concentrations of selected halogenated gases over time (Figure 4).

• Global atmospheric concentrations of ozone over time (Figure 5).

2. Revision History

April 2010: Indicator published.

December 2012: Updated indicator with data through 2011. Added nitrogen trifluoride to Figure 4.
August 2013: Updated indicator with data through 2012.

May 2014: Updated Figures 1, 2, and 3 with data through 2013, and Figure 4 with data through

2012. Added Figure 5 to show trends in ozone.

June 2015: Updated Figures 1, 2, and 3 with data through 2014.

April 2016: Updated Figure 1 with data through 2015.

August 2016: Updated Figures 2, 3, and 4 with data through 2015; updated Figure 5 with data

through 2014.

April 2021: Updated Figures 1, 2, and 3 with data through 2019; updated Figures 4 and 5 with

data through 2018.

Data Sources

3. Data Sources

Ambient concentration data used to develop this indicator were taken from the following sources:

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
Figure 1. Global Atmospheric Concentrations of Carbon Dioxide Over Time

•	Antarctic ice cores: approximately 803,719 BCE to 2001 CE—Bereiter et al. (2015). This is a
composite of the following individual sources (see Bereiter et al2015, for corresponding
full citations):

¦	Law Dome (Rubino et al., 2013)

¦	Law Dome (MacFarling Meure et al., 2006)

¦	Dome C (Monnin et al., 2001, 2004)

¦	WAIS (Marcott et al., 2014) minus 4 parts per million by volume

¦	Siple Dome (Ahn et al., 2014)

¦	TALDICE (Bereiter et al., 2012)

¦	EDML (Bereiter et al., 2012)

¦	Dome C Sublimation (Schneider et al., 2013)

¦	Vostok (Petit et al., 1999)

¦	Dome C (Siegenthaler et al., 2005)

¦	Dome C (Bereiter et al., 2014)

•	Mauna Loa, Hawaii: 1959 CE to 2019 CE—NOAA (2020a).

•	Barrow, Alaska: 1974 CE to 2019 CE; Cape Matatula, American Samoa: 1976 CE to 2019 CE;
South Pole, Antarctica: 1976 CE to 2019 CE—NOAA (2020b).

•	Cape Grim, Australia: 1977 CE to 2019 CE—CSIRO (2020a).

•	Shetland Islands, Scotland: 1993 CE to 2002 CE—Steele et al. (2007).

•	Lampedusa Island, Italy: 1993 CE to 2000 CE—Chamard et al. (2001).

Figure 2. Global Atmospheric Concentrations of Methane Over Time

•	EPICA Dome C, Antarctica: approximately 797,446 BCE to 1937 CE—Loulergue et al. (2008).

•	Law Dome, Antarctica: approximately 1008 CE to 1980 CE—Etheridge et al. (2002).

•	Cape Grim, Australia: 1985 CE to 2019 CE—CSIRO (2020b).

•	Mauna Loa, Hawaii: 1984 CE to 2019 CE—NOAA (2020c).

•	Shetland Islands, Scotland: 1993 CE to 2001 CE—Steele et al. (2002).

Figure 3. Global Atmospheric Concentrations of Nitrous Oxide Over Time

•	EPICA Dome C, Antarctica: approximately 796,475 BCE to 1937 CE—Schilt et al. (2010).

•	Antarctica: approximately 1903 CE to 1976 CE—Battle et al. (1996).

•	Cape Grim, Australia: 1979 CE to 2019 CE—CSIRO (2020c).

•	South Pole, Antarctica: 1998 CE to 2019 CE; Barrow, Alaska: 1999 CE to 2019 CE; Mauna Loa,
Hawaii: 2000 CE to 2019 CE—NOAA (2020d).

Figure 4. Global Atmospheric Concentrations of Selected Halogenated Gases, 1978-2018

Global average atmospheric concentration data for selected halogenated gases were obtained from the
Advanced Global Atmospheric Gases Experiment (AGAGE, 2019) for all species. Historical data for some
species, whose time series could be extended by including data from other studies, were obtained from
these sources:

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

2

-------
• The National Oceanic and Atmospheric Administration (NOAA, 2019) for halon-1211 before
January 2004.

• Rigby (2017) for nitrogen trifluoride data before February 2015 and PFC-14 data before April
2006.

• Historical data provided by Dr. Ray Wang of AGAGE, representing AGAGE-approved data
that predate the period currently available for download on the AGAGE website (AGAGE,
2011).

A similar figure based on AGAGE and NOAA data appears in the Intergovernmental Panel on Climate
Change's (IPCC's) Fifth Assessment Report (see Figure 2.4 in IPCC, 2013).

Figure 5. Global Atmospheric Concentrations of Ozone, 1979-2018

Ozone data were obtained from several National Aeronautics and Space Administration (NASA) sources:

• The Solar Backscatter Ultraviolet (SBUV) merged ozone data set (NASA, 2019) for total
ozone.

• The Tropospheric Ozone Residual (TOR) (NASA, 2013) and Ozone Monitoring Instrument
(OMI) Level 2 (NASA, 2020) data sets for tropospheric ozone.

4. Data Availability

The data used to develop Figures 1, 2, 3, and 5 of this indicator are publicly available and can be
accessed from the references listed in Section 3. There are no known confidentiality issues.

Data for all of the halogenated gases in Figure 4 were downloaded from the AGAGE website at:
http://agage.eas.gatech.edu/data archive/global mean. Additional historical monthly data for some of
these gases were provided in spreadsheet form by Dr. Ray Wang of the AGAGE project team (AGAGE,
2011). Historical bimonthly data for halon-1211 were provided in spreadsheet form by Dr. Stephen
Montzka of NOAA (NOAA, 2019). NOAA's website (www.esrl.noaa.gov/gmd/hats) also provides access
to the global averages that are shown in Figure 4. Historical data for nitrogen trifluoride are based on
measurements that were originally published in Arnold et al. (2013) and subsequently updated by the
lead author and other collaborators. Historical PFC-14 and nitrogen trifluoride data were provided in
spreadsheet form by Matthew Rigby of the University of Bristol. AGAGE's database includes
measurements for halon-1211 and nitrogen trifluoride starting in April 2006 and February 2015,
respectively.

Methodology

5. Data Collection

This indicator shows trends in atmospheric concentrations of several major GHGs that enter the
atmosphere at least in part because of human activities: carbon dioxide (C02), methane (CH4), nitrous
oxide (N20), selected halogenated gases, and ozone.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
Figures 1, 2, 3, and 4. Global Atmospheric Concentrations of Carbon Dioxide, Methane, Nitrous Oxide,
and Selected Halogenated Gases Over Time

Figures 1, 2, 3, and 4 aggregate comparable, high-quality data from individual studies that each focused
on different locations and time frames. Data since the mid-20th century come from global networks that
use standard monitoring techniques to measure the concentrations of gases in the atmosphere. Older
measurements of atmospheric concentrations come from ice cores—specifically, measurements of gas
concentrations in air bubbles that were trapped in ice at the time the ice was formed. Scientists have
spent years developing and refining methods of measuring gases in ice cores as well as methods of
dating the corresponding layers of ice to determine their age. Ice core measurements are a widely used
method of reconstructing the composition of the atmosphere before the advent of direct monitoring
techniques.

This indicator presents a compilation of data generated by numerous sampling programs. The citations
listed in Section 3 describe the specific approaches taken by each program. Gases are measured by mole
fraction relative to dry air.

C02, CH4, N20, and most of the halogenated gases presented in this indicator are considered to be well-
mixed globally, due in large part to their long residence times in the atmosphere. Thus, while
measurements over geological time tend to be available only for regions where ice cores can be
collected (e.g., the Arctic and Antarctic regions), these measurements are believed to adequately
represent concentrations worldwide. Recent monitoring data have been collected from a greater variety
of locations, and the results show that concentrations and trends are indeed very similar throughout the
world, although relatively small variations can be apparent across different locations.

Most of the gases shown in Figure 4 have been measured around the world numerous times per year.
One exception is nitrogen trifluoride, which is an emerging gas of concern for which measurements have
only recently started to become more widespread. The curve for nitrogen trifluoride in Figure 4 is based
on samples collected in Australia and California through 2010, plus samples collected approximately
monthly in California since 2010 and at additional sites in the most recent years. Measurements of air
samples collected before the mid-1990s are included in the data set, but they do not appear in Figure 4
because they are below the minimum concentration shown on the y-axis.

Nitrogen trifluoride was measured by the Medusa gas chromatography with mass spectrometry (GCMS)
system, with refinements described in Weiss et al. (2008), Arnold et al. (2012), and Arnold et al. (2013).
Mole fractions of the other halogenated gases were collected by AGAGE's Medusa GCMS system, or
similar methods employed by NOAA.

Figure 5. Global Atmospheric Concentrations of Ozone, 1979-2018

Unlike the gases in Figures 1, 2, 3, and 4, which are measured as atmospheric concentrations near
ground level, Figure 5 describes the total "thickness" of ozone in the Earth's atmosphere. This
measurement is called total column ozone, and it is typically measured in Dobson units. One Dobson
unit represents a layer of gas that would be 10 micrometers (nm) thick under standard temperature and
pressure (0°C/32°F and 0.987 atmospheres of air pressure).

Atmospheric ozone concentrations for this indicator are based on measurements by three sets of
satellite instruments:

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
• SBUV. The SBUV observing system consists of a series of instruments aboard nine satellites
that have collectively covered the period from 1970 to present, except for a gap from 1972
to 1978. The SBUV measures the total ozone profile from the Earth's surface to the upper
edge of the atmosphere (total column ozone) by analyzing solar backscatter radiation, which
is the visible light and ultraviolet radiation that the Earth's atmosphere reflects back to
space. This instrument can be used to determine the amount of ozone in each of 21 discrete
layers of the atmosphere, which are then added together to get total column ozone. For a
table of specific SBUV satellite instruments and the time periods they cover, see:
http://acdb-ext.gsfc.nasa.gov/Data services/merged/instruments.html. A new instrument,
the Ozone Mapping Profiler Suite (OMPS) Nadir Profiler, will continue the SBUV series.
Although instrument design has improved over time, the basic principles of the
measurement technique and processing algorithm remain the same, lending consistency to
the record. For more information about the SBUV data set and how it was collected, see
McPeters et al. (2013) and the references listed at: http://acdb-
ext.gsfc.nasa.gov/Data services/merged/index.html.

• Total Ozone Mapping Spectrometer (TOMS). TOMS instruments have flown on four
satellite missions that collectively cover the period from 1978 to 2005, with the exception of
a period from late 1994 to early 1996 when no TOMS instrument was in orbit. Like the
SBUV, the TOMS measured total ozone in the Earth's atmosphere by analyzing solar
backscatter radiation. For more information about TOMS missions and instrumentation, see:
https://eospso.nasa.gov/missions/total-ozone-mapping-spectrometer-earth-probe.

• Aura OMI and Microwave Limb Sounder (MLS). The Aura satellite was launched in 2004,
and its instruments (including OMI and MLS) were still collecting data as of 2021. The OMI
instrument measures total column ozone by analyzing solar backscatter radiation. In
contrast, the MLS measures emissions of microwave radiation from the Earth's atmosphere.
This method allows the MLS to characterize the temperature and composition of specific
layers of the atmosphere, including the amount of ozone within the stratosphere. To learn
more about the Aura mission and its instruments, visit: https://aura.gsfc.nasa.gov and:
https://aura.gsfc.nasa.gov/scinst.html.

The instruments described above have flown on polar-orbiting satellites, which collect measurements
that cover the entire surface of the Earth. For reasons of accuracy described in Section 9, however, this
indicator is limited to data collected between 50°N and 50°S latitude. Solar backscatter measurements
are restricted to daytime, when the sun is shining on a particular part of the Earth and not too low in the
sky (i.e., avoiding measurements near sunrise or sunset).

6. Indicator Derivation

EPA obtained and compiled data from various GHG measurement programs and plotted these data in
graphs. No attempt was made to project concentrations backward before the beginning of the ice core
record (or the start of monitoring, in the case of Figures 4 and 5) or forward into the future.

Figures 1, 2, and 3. Global Atmospheric Concentrations of Carbon Dioxide, Methane, and Nitrous Oxide
Over Time

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
Figures 1, 2, and 3 plot data at annual or multi-year intervals; with ice cores, consecutive data points are
often spaced many years apart. EPA used the data exactly as reported by the organizations that
collected them, with the following exceptions:

• Some of the recent time series for C02, CH4, and N20 consisted of monthly measurements.
EPA averaged these monthly measurements to arrive at annual values to plot in the graphs.
A few years did not have data for all 12 months. If at least 10 months of data were present
in a given year, EPA averaged the available data to arrive at an annual value. If fewer than
10 monthly measurements were available, that year was excluded from the graph.

• Some ice core records were reported in terms of the age of the sample or the number of
years before present. EPA converted these dates into calendar years.

• A few ice core records had multiple values at the same point in time (i.e., two or more
different measurements for the same year). These values were generally comparable and
never varied by more than 4.8 percent. In such cases, EPA averaged the values to arrive at a
single atmospheric concentration per year.

Figures 1, 2, and 3 present a separate line for each data series or location where measurements were
collected. No methods were used to portray data for locations other than where measurements were
made. The indicator does imply, however, that the values in the graphs represent global atmospheric
concentrations—an appropriate assumption because the gases covered by this indicator have long
residence times in the atmosphere and are considered to be well-mixed. In the indicator text, the key
points refer to the concentration for the most recent year available. If data were available for more than
one location, the text refers to the average concentration across these locations.

Figure 4. Global Atmospheric Concentrations of Selected Halogenated Gases, 1978-2018

Figure 4 plots data at sub-annual intervals (i.e., several data points per year). EPA used the data exactly
as reported by the organizations that collected them. Figure 4 presents one trend line for each
halogenated gas, and these lines represent average concentrations across all measurement sites
worldwide. These data represent monthly average mole fractions for each species, with two exceptions:
halon-1211 data are only available at two-month intervals for data before January 2004, and nitrogen
trifluoride measurements were converted into global annual average mole fractions using a model
described in Arnold et al. (2013). The 2014 update of nitrogen trifluoride represented a change from the
version of this indicator that EPA initially published in December 2012. At the time of the December
2012 version, modeled global annual average mole fractions had not yet been published in the
literature, so EPA's indicator instead relied upon individual measurements of nitrogen trifluoride that
were limited to the Northern Hemisphere.

EPA's 2021 web update represented a change from previous versions of this indicator by showing data
from the AGAGE project for all the gas species presented. Previous versions presented data from other
sources for all years of halon-1211, nitrogen trifluoride, and PFC-14. Beginning with the 2021 update,
this indicator presents AGAGE data for all newly updated values, while preserving the historical data
values presented in previous versions of the indicator, which predate a species' inclusion in the AGAGE
data set. In the case of all three gases, this switch resulted in the replacement of values shown in
previous versions of this indicator for dates that overlapped with newly obtained data from AGAGE.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
Data are available for additional halogenated species, but to make the most efficient use of the space
available, EPA selected a subset of gases that are relatively common, have several years of data
available, show marked growth trends (either positive or negative), and/or collectively represent most
of the major categories of halogenated gases. The inclusion of nitrogen trifluoride here is based on
several factors. Like perfluoromethane (PFC-14 or CF4), perfluoroethane (PFC-116 or C2F6), and sulfur
hexafluoride, nitrogen trifluoride is a widely produced, fully fluorinated gas with a very high 100-year
global warming potential (17,200) and a long atmospheric lifetime (740 years). Nitrogen trifluoride has
experienced a rapid increase in emissions (i.e., more than 10 percent per year) due to its use in
manufacturing semiconductors, flat screen displays, and thin film solar cells. It began to replace
perfluoroethane in the electronics industry in the late 1990s.

To examine the possible influence of phase-out and substitution activities under the Montreal Protocol
on Substances That Deplete the Ozone Layer, EPA divided Figure 4 into two panels: one for substances
officially designated as "ozone-depleting" and one for all other halogenated gases.

Figure 5. Global Atmospheric Concentrations of Ozone, 1979-2018

NASA converted the satellite measurements into meaningful data products using the following methods:

• Data from all SBUV instruments were processed using the Version 8.6 algorithm (Bhartia et
al., 2012; Kramarova et al., 2013b). The resulting data set indicates the amount of ozone in
each of 21 distinct atmospheric layers, in Dobson units.

• NASA developed the TOR data set, which represents ozone in the troposphere only. They
did so by starting with total column ozone measurements from TOMS and SBUV, then
subtracting the portion that could be attributed to the stratosphere. NASA developed this
method using information about the height of the tropopause (the boundary between the
troposphere and the stratosphere) over time and space, stratosphere-only ozone
measurements from the Stratospheric Aerosol and Gas Experiment (SAGE) instrument that
flew on some of the same satellites as TOMS, analysis of larger-scale patterns in
stratospheric ozone distribution, and empirical corrections based on field studies. These
methods are described in detail at: https://science-data.larc.nasa.gov/TOR/data.html and in
Fishman et al. (2003) and the references cited therein.

• NASA developed the OMI Level 2 tropospheric ozone data set by essentially subtracting MLS
stratospheric ozone observations from concurrent OMI total column ozone observations.
Ziemke et al. (2006) describe these methods in more detail.

EPA performed the following additional processing steps to convert NASA's data products into an easy-
to-understand indicator:

• EPA obtained SBUV data in the form of monthly averages for each layer of the atmosphere
(total: 21 layers) by latitude band (i.e., average ozone levels for each 5-degree band of
latitude). For each latitude band, EPA added the ozone levels for NASA's 21 atmospheric
layers together to get total column ozone. Next, because each latitude band represents a
different amount of surface area of the atmosphere (for example the band near the North
Pole from 85°N to 90°N covers a much smaller surface area than the band near the equator
from 0° to 5°N), EPA calculated a global average using cosine area weighting. The global

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
average in this indicator only covers the latitude bands between 50°N and 50°S for
consistency of satellite coverage. EPA then combined the monthly averages to obtain annual
averages.

• EPA obtained TOR and OMI Level 2 data as a grid of monthly average tropospheric ozone
levels. Both data sets are divided into grid cells measuring 1-degree latitude by 1.25-degrees
longitude and are only available between 50°N and 50°S. EPA calculated global monthly
averages for each 1-degree latitude band by averaging over all grid cells in that band, then
used cosine area weighting to calculate an average for the entire range from 50°N to 50°S.
EPA combined the monthly averages to obtain annual averages.

In Figure 5, the "total column" line comes from the SBUV data set. Because of missing data from mid-
1972 through late 1978, EPA elected to start the graph at 1979. From 1979 to present, all years have
complete SBUV data.

The "troposphere" line in Figure 5 is based on the TOR data set from 1979 to 2004, the OMI Level 2 data
set from 2006 to present, and an average of TOR and OMI Level 2 data for 2005. To correct for
differences between the two instruments, EPA adjusted all OMI data points upward by 1.799 Dobson
units, which is the documented difference during periods of overlap in 2004. This is a standard
bootstrapping approach. Data are not shown from 1994 to 1996 because no TOMS instrument was in
orbit from late 1994 to early 1996, so it was not possible to calculate annual averages from the TOR data
set during these three years. The "stratosphere" line in Figure 5 was calculated by subtracting the
"troposphere" series from the "total column" series.

Indicator Development

Figures 1, 2, 3, and 4 were first published as part of EPA's 2010 and 2012 climate change indicator
reports. EPA added Figure 5 for the 2014 edition to address one of the key limitations of the previous
indicator and to reflect the scientific community's growing awareness of the importance of tropospheric
ozone as a contributor to climate change.

Scientists measure the amount of ozone in the atmosphere using two complementary methods. In
addition to NASA's satellite-based data collection, NOAA operates a set of ground-based sites using
devices called Dobson ozone spectrophotometers, which point upward and measure total column ozone
on clear days. A set of 10 of these sites constitute the NOAA Ozone Reference Network. Measurements
have been collected at some of these sites since the 1920s, and the resulting data are available at:
www.esrl.noaa.gov/gmd/ozwv/dobson.

When developing this indicator, EPA chose to focus on satellite measurements because they allow total
column ozone to be separated into tropospheric and stratospheric components, which facilitates greater
understanding of the complex roles that ozone, ozone-depleting substances, and emissions of ozone
precursors play in climate change. In addition to the fact that satellite-based data products were readily
available to assess ozone concentrations by layer of the atmosphere, tropospheric ozone is short-lived
and not globally mixed, so satellite-based measurements arguably provide more complete coverage of
this greenhouse gas than a finite set of ground-based stations. Nonetheless, as described in Section 7,
NOAA's ground-based measurements still play an important role in this indicator because they provide
independent verification of the trends detected by satellites.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
7. Quality Assurance and Quality Control

The data for this indicator have generally been taken from carefully constructed, peer-reviewed studies.
Quality assurance and quality control procedures are addressed in the individual studies, which are cited
in Section 3. Additional documentation of these procedures can be obtained by consulting with the
principal investigators who developed each of the data sets.

NASA selected SBUV data for their official merged ozone data set based on the results of a detailed
analysis of instrumental uncertainties (DeLand et al., 2012) and comparisons against independent
satellite and ground-based profile observations (Kramarova et al., 2013b; Labow et al., 2013). NASA
screened SBUV data using the following rules:

• Data from the SBUV/2 instrument on the NOAA-9 satellite are not included due to multiple
instrumental issues (DeLand et al., 2012).

• Only measurements made between 8 AM and 4 PM Equatorial Crossing Time are included in
the merged satellite ozone data set, with one exception in 1994-1995, when NOAA 11 data
were included to avoid a gap in the data.

• When data from more than one SBUV instrument are available, NASA used a simple average
of the data.

• Data were filtered for aerosol contamination after the eruptions of El Chichon (1982) and
Mt. Pinatubo (1991).

Satellite data have been validated against ground-based measurements from NOAA's Ozone Reference
Network. Figure TD-1 below shows how closely these two complementary data sources track each other
over time.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
Figure TD-1. Yearly Average Change in Global Total Column Ozone Since 1979

NOAA Ozone Reference
Network (ground-based)
NASA SBUV dataset (satellite)

1975 1980 1985 1990

1995
Year

2000 2005 2010 2015

Analysis

8. Comparability Over Time and Space

Data have been collected using a variety of methods over time and space, but these methodological
differences are expected to have little bearing on the overall conclusions for this indicator. The
concordance of trends among multiple data sets collected using different program designs provides
some assurance that the trends depicted actually represent changes in atmospheric conditions, rather
than some artifact of sampling design.

Figures 1, 2, 3, and 4. Global Atmospheric Concentrations of Carbon Dioxide, Methane¦, Nitrous Oxide,
and Selected Halogenated Gases

The gases covered in Figures 1, 2, 3, and 4 are all long-lived GHGs that are relatively evenly distributed
globally. Thus, measurements collected at one particular location have been shown to be representative
of average concentrations worldwide. Three gas species have measurements from more than one data
source that have been combined into a single line on the graph (halon-1211, nitrogen trifluoride, and
PFC-14). The switch from one data source to another occurs at a different time for each of the three
species. In the case of halon-1211, the transition also represents a change from bimonthly to monthly
measurements. A comparison of data where there was temporal overlap of readings between AGAGE
(2019) and NOAA (2019) or Rigby (2017) indicated that there was not more than a 2 percent difference
on average between the sources.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
Figure 5. Global Atmospheric Concentrations of Ozone, 1979-2018

Because ozone concentrations vary over time and space, Figure 5 uses data from satellites that cover
virtually the entire globe, and the figure shows area-weighted global averages. These satellite data have
undergone extensive testing to identify errors and biases through comparison with independent satellite
and ground-based profile observations, including the NOAA reference ozone network (Kramarova et al.,
2013b).

9. Data Limitations

Factors that may impact the confidence, application, or conclusions drawn from this indicator are as
follows:

1. This indicator does not track water vapor because of its spatial and temporal variability. Human
activities have only a small direct impact on water vapor concentrations, but there are
indications that increasing global temperatures are leading to increasing levels of atmospheric
humidity (Dai et al., 2011).

2. Some radiatively important atmospheric constituents that are substantially affected by human
activities (such as black carbon, aerosols, and sulfates) are not included in this indicator because
of their spatial and temporal variability.

3. This indicator includes several of the most important halogenated gases, but some others are
not shown. Many other halogenated gases are also GHGs, but Figure 4 is limited to a set of
common examples that represent most of the major types of these gases.

4. Ice core measurements are not taken in real time, which introduces some error into the date of
the sample. Dating accuracy for the ice cores ranges up to plus or minus 20 years (often less),
depending on the method used and the time period of the sample. Diffusion of gases from the
samples, which would tend to reduce the measured values, could also add a small amount of
uncertainty.

5. Factors that could affect satellite-based ozone measurements include orbital drift, instrument
differences, and solar zenith angle (the angle of incoming sunlight) at the time of measurement.
As discussed in Section 10, however, the data have been filtered and calibrated to account for
these factors. For example, Figure 5 has been restricted to the zone between 50°N and 50°S
latitude because at higher latitudes the solar zenith angles would introduce greater uncertainty
and because the lack of sunlight during winter months creates data gaps.

10. Sources of Uncertainty

Figures 1, 2, 3, and 4. Global Atmospheric Concentrations of Carbon Dioxide, Methane¦, Nitrous Oxide,
and Selected Halogenated Gases

Direct measurements of atmospheric concentrations, which cover approximately the last 50 years, are
of a known and high quality. Generally, standard errors and accuracy measurements are computed for
the data.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
For ice core measurements, uncertainties result from the actual gas measurements as well as the dating
of each sample. Uncertainties associated with the measurements are believed to be relatively small,
although diffusion of gases from the samples might also add to the measurement uncertainty. Dating
accuracy for the ice cores is believed to be within plus or minus 20 years, depending on the method
used and the time period of the sample. This level of uncertainty is insignificant, however, considering
that some ice cores characterize atmospheric conditions for time frames of hundreds of thousands of
years. The original scientific publications (see Section 3) provide more detailed information on the
estimated uncertainty within the individual data sets.

Visit the Carbon Dioxide Information Analysis Center (CDIAC) website (https://cdiac.ess-
dive.lbl.gov/tracegases.html) for more information on the accuracy of both direct and ice core
measurements.

Overall, the concentration increase in GHGs in the past century is far greater than the estimated
uncertainty of the underlying measurement methodologies. It is highly unlikely that the concentration
trends depicted in this set of figures are artifacts of uncertainty.

Figure 5. Global Atmospheric Concentrations of Ozone, 1979-2018

NASA has estimated uncertainties for the merged SBUV satellite ozone data set, mostly focusing on
errors that might affect trend analysis. Constant offsets and random errors will make no difference in
the trend, but smoothing error and instrumental drift can potentially affect trend estimates. The
following discussion describes these sources of error, the extent to which they affect the data used for
this indicator, and steps taken to minimize the corresponding uncertainty.

• The main source of error in the SBUV data is a smoothing error due to profile variability that
the SBUV observing system cannot inherently measure (Bhartia et al., 2012; Kramarova et
al., 2013a). NASA's SBUV data set is divided into 21 distinct layers of the atmosphere, and
the size of the smoothing error varies depending on the layer. For the layers that make up
most of the stratosphere (specifically between 16 hectopascals [hPa] and 1 hPa of pressure
in the tropics [20°S to 20°N] and 25 hPa to 1 hPa outside the tropics), the smoothing error
for the SBUV monthly mean profiles is approximately 1 percent, indicating that SBUV data
are capable of accurately representing ozone changes in this part of the atmosphere. For
the SBUV layers that cover the troposphere, the lower stratosphere, and above the
stratosphere (air pressure less than 1 hPa), the smoothing errors are larger: up to 8 to 15
percent. The influence of these smoothing errors has been minimized by adding all of the
individual SBUV layers together to examine total column ozone.

• Long-term drift can only be estimated through comparison with independent data sources.
NASA validated the SBUV merged ozone data set against independent satellite observations
and found that drifts are less than 0.3 percent per year and mostly insignificant.

• Several SBUV instruments have been used over time, and each instrument has specific
characteristics. NASA estimated the offsets between pairs of SBUV instruments when they
overlap (DeLand et al., 2012) and found that mean differences are within 7 percent, with
corresponding standard deviations of 1 to 3 percent. The SBUV Version 8.6 algorithm adjusts
for these differences based on a precise comparison of radiance measurements during
overlap periods.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
• Because the SBUV instruments use wavelengths that have high sensitivity to ozone, the total
column ozone calculated from this method is estimated to have a 1 to 2 Dobson unit
accuracy for solar zenith angles up to 70 degrees—i.e., when the sun is more than 20
degrees above the horizon (Bhartia et al., 2012). Measurements taken when the sun is lower
in the sky have less accuracy, which is why the SBUV data in this indicator have been mostly
limited to measurements made between 8 AM and 4 PM Equatorial Crossing Time, and one
reason why the data have been limited to the area between 50°N and 50°S latitude (77
percent of the Earth's surface area).

Fishman et al. (2003) describe uncertainties in the TOR tropospheric ozone data set. Calculations of
global average TOR may vary by up to 5 Dobson units, depending on which release of satellite data is
used. For information about uncertainty in the OMI Level 2 tropospheric ozone data set, see Ziemke et
al. (2006), which describes in detail how OMI data have been validated against ozonesonde data. Both
of these data sets have been limited to the zone between 50°N and 50°S latitude because of the solar
angle limitation described above. Based on the considerations, adjustment steps, and validation steps
described above, it is unlikely that the patterns depicted in Figure 5 are artifacts of uncertainty.

11. Sources of Variability

Figures 1, 2, 3, and 4. Global Atmospheric Concentrations of Carbon Dioxide, Methane, Nitrous Oxide,
and Selected Halogenated Gases

Atmospheric concentrations of the long-lived GHGs vary with both time and space. The data presented
in this indicator, however, have extraordinary temporal coverage. For carbon dioxide, methane, and
nitrous oxide, concentration data span several hundred thousand years; for the halogenated gases, data
span virtually the entire period during which these largely synthetic gases were widely used. While
spatial coverage of monitoring stations is more limited, most of the GHGs presented in this indicator are
considered to be well-mixed globally, due in large part to their long residence times in the atmosphere.

Figure 5. Global Atmospheric Concentrations of Ozone, 1979-2018

Unlike the other gases described in this indicator, ozone is relatively short-lived in the troposphere, with
a typical lifetime of only a few weeks. Concentrations of both tropospheric and stratospheric ozone vary
spatially at any given time; for example, Fishman et al. (2003) use the TOR to show noticeably elevated
levels of tropospheric ozone over heavily populated and industrialized regions. Fishman et al. (2003) also
show seasonal variations. This indicator accounts for both spatial and temporal variations by presenting
global annual averages.

12. Statistical/Trend Analysis

This indicator presents a time series of atmospheric concentrations of GHGs. For the long-lived gases, no
statistical techniques or analyses have been used here to characterize the long-term trends or their
statistical significance. The latest authoritative scientific assessments have concluded that
concentrations of these gases substantially exceed the highest concentrations recorded in ice cores
during the past 800,000 years (IPCC, 2013).

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

-------
For ozone, EPA used ordinary least-squares linear regressions as a screening-level assessment of
whether changes in ozone levels over time have been statistically significant. This analysis yielded the
following results:

•	A regression of the total column ozone data from 1979 to 2018 shows a significant decrease
of approximately 0.13 Dobson units per year (p = 0.001).

•	Further analysis of the total column ozone data shows a rapid decline over the first decade
and a half of the data record (1979-1994), with insignificant change after that. A regression
analysis for 1979-1994 shows a significant decline of about 0.68 Dobson units per year (p <
0.001), while the regression for the remainder of the data record (1995-2018) shows an
insignificant change (p = 0.48).

•	A regression of tropospheric ozone from 1979 to 2018 shows a significant increase of 0.08
Dobson units per year (p < 0.001).

References

AGAGE (Advanced Global Atmospheric Gases Experiment). 2011. Data provided to ERG (an EPA
contractor) by Ray Wang, Georgia Institute of Technology. November 2011.

AGAGE (Advanced Global Atmospheric Gases Experiment). 2019. ALE/GAGE/AGAGE database. Updated
January 8, 2019. Accessed December 2020. http://agage.mit.edu.

Arnold, T., J. Muhle, P.K. Salameh, C.M. Harth, D.J. Ivy, and R.F. Weiss. 2012. Automated measurement
of nitrogen trifluoride in ambient air. Anal. Chem. 84(ll):4798-4804.

Arnold, T., C.M. Harth, J. Muhle, A.J. Manning, P.K. Salameh, J. Kim, D.J. Ivy, L.P. Steele, V.V. Petrenko,
J.P. Severinghaus, D. Baggenstos, and R.F. Weiss. 2013. Nitrogen trifluoride global emissions estimated
from updated atmospheric measurements. P. Natl. Acad. Sci. USA 110(6):2029-2034.

Battle, M., M. Bender, T. Sowers, P. Tans, J. Butler, J. Elkins, J. Ellis, T. Conway, N. Zhang, P. Lang, and A.
Clarke. 1996. Atmospheric gas concentrations over the past century measured in air from firn at the
South Pole. Nature 383:231-235. https://daac.ornl.gov/cgi-bin/dsviewer.plPds id=797.

Bereiter, B., S. Eggleston, J. Schmitt, C. Nehrbass-Ahles, T.F. Stocker, H. Fischer, S. Kipfstuhl, and J.
Chappellaz. 2015. Revision of the EPICA Dome C C02 record from 800 to 600 kyr before present.
Geophys. Res. Let. 42(2):542-549. www.ncdc.noaa.gov/paleo-search/studv/17975.

Bhartia, P.K., R.D. McPeters, L.E. Flynn, S. Taylor, N.A. Kramarova, S. Frith, B. Fisher, and M. DeLand.
2012. Solar Backscatter UV (SBUV) total ozone and profile algorithm. Atmos. Meas. Tech. Discuss.
5:5913-5951.

Chamard, P., L. Ciattaglia, A. di Sarra, and F. Monteleone. 2001. Atmospheric carbon dioxide record from
flask measurements at Lampedusa Island. In: Trends: A compendium of data on global change. Oak

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

14

-------
Ridge, TN: U.S. Department of Energy. Accessed September 14, 2005. http://cdiac.ess-
dive.lbl.gov/trends/co2/lampis.html.

CSIRO (Commonwealth Scientific and Industrial Research Organisation). 2020a. Monthly mean baseline
(background) carbon dioxide concentrations measured at the Cape Grim Baseline Air Pollution Station,
Tasmania, Australia. Updated December 2020. Accessed December 29, 2020.
http://capegrim.csiro.au/GreenhouseGas/data/CapeGrim C02 data download.csv.

CSIRO (Commonwealth Scientific and Industrial Research Organisation). 2020b. Monthly mean baseline
(background) methane concentrations measured at the Cape Grim Baseline Air Pollution Station,
Tasmania, Australia. Updated December 2020. Accessed December 29, 2020.
http://capegrim.csiro.au/GreenhouseGas/data/CapeGrim CH4 data download.csv.

CSIRO (Commonwealth Scientific and Industrial Research Organisation). 2020c. Monthly mean baseline
(background) nitrous oxide concentrations measured at the Cape Grim Baseline Air Pollution Station,
Tasmania, Australia. Updated December 2020. Accessed December 29, 2020.
http://capegrim.csiro.au/GreenhouseGas/data/CapeGrim N2Q data download.csv.

Dai, A., J. Wang, P.W. Thorne, D.E. Parker, L. Haimberger, and X.L. Wang. 2011. A new approach to
homogenize daily radiosonde humidity data. J. Climate 24(4):965-991.
http://iournals.ametsoc.Org/doi/abs/10.1175/2010JCLI3816.l.

DeLand, M.T., S.L. Taylor, L.K. Huang, and B.L. Fisher. 2012. Calibration of the SBUV version 8.6 ozone
data product. Atmos. Meas. Tech. Discuss. 5:5151-5203.

Etheridge, D.M., L.P. Steele, R.J. Francey, and R.L. Langenfelds. 2002. Historic CH4 records from Antarctic
and Greenland ice cores, Antarctic firn data, and archived air samples from Cape Grim, Tasmania. In:
Trends: A compendium of data on global change. Oak Ridge, TN: U.S. Department of Energy. Accessed
September 13, 2005. http://cdiac.ess-dive.lbl.gov/trends/atm meth/lawdome meth.html.

Fishman, J., A.E. Wozniak, and J.K. Creilson. 2003. Global distribution of tropospheric ozone from
satellite measurements using the empirically corrected tropospheric ozone residual technique:
Identification of the regional aspects of air pollution. Atmos. Chem. Phys. 3:893-907.

IPCC (Intergovernmental Panel on Climate Change). 2013. Climate change 2013: The physical science
basis. Working Group I contribution to the IPCC Fifth Assessment Report. Cambridge, United Kingdom:
Cambridge University Press, www.ipcc.ch/report/ar5/wgl.

Kramarova, N.A., P.K. Bhartia, S.M. Frith, R.D. McPeters, and R.S. Stolarski. 2013a. Interpreting SBUV
smoothing errors: An example using the quasi-biennial oscillation. Atmos. Meas. Tech. 6:2089-2099.

Kramarova, N.A., S.M. Frith, P.K. Bhartia, R.D. McPeters, S.L. Taylor, B.L. Fisher, G.J. Labow, and M.T.
DeLand. 2013b. Validation of ozone monthly zonal mean profiles obtained from the version 8.6 Solar
Backscatter Ultraviolet algorithm. Atmos. Chem. Phys. 13:6887-6905.

Labow, G.J., R.D. McPeters, P.K. Bhartia, and N.A. Kramarova. 2013. A comparison of 40 years of SBUV
measurements of column ozone with data from the Dobson/Brewer Network. J. Geophys. Res. Atmos.
118(13):7370-7378.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

15

-------
Loulergue, L., A. Schilt, R. Spahni, V. Masson-Delmotte, T. Blunier, B. Lemieux, J.-M. Barnola, D. Raynaud,
T.F. Stocker, and J. Chappellaz. 2008. Orbital and millennial-scale features of atmospheric CH4 over the
past 800,000 years. Nature 453:383-386. www.ncdc.noaa.gov/paleo-search/studv/6093.

McPeters, R.D., P.K. Bhartia, D. Haffner, G.J. Labow, and L. Flynn. 2013. The v8.6 SBUV ozone data
record: An overview. J. Geophys. Res. Atmos. 118(14):8032-8039.

NASA (National Aeronautics and Space Administration). 2013. Data—TOMS/SBUV TOR data products.
Accessed November 2013. https://science-data.larc.nasa.gov/TOR/data.html.

NASA (National Aeronautics and Space Administration). 2019. SBUV merged ozone dataset (MOD).
Version 8.6. Updated November 20, 2019. Accessed December 2020. http://acdb-
ext.gsfc.nasa.gov/Data services/merged/index.html.

NASA (National Aeronautics and Space Administration). 2020. Tropospheric ozone data from AURA
OMI/MLS. Accessed December 2020. http://acdb-
ext.gsfc.nasa.gov/Data services/cloud slice/new data.html.

NOAA (National Oceanic and Atmospheric Administration). 2019. Halocarbons and Other Atmospheric
Trace Species group (HATS). Updated October 2019. Accessed January 2021.
ftp://ftp.cmdl.noaa.gov/hats/Total CI Br.

NOAA (National Oceanic and Atmospheric Administration). 2020a. Annual mean carbon dioxide
concentrations for Mauna Loa, Hawaii. Updated September 23, 2020. Accessed December 2020.
www.esrl.noaa.gov/gmd/ccgg/trends/data.html.

NOAA (National Oceanic and Atmospheric Administration). 2020b. Monthly mean carbon dioxide
concentrations for Barrow, Alaska; Cape Matatula, American Samoa; and the South Pole. Updated
August 26, 2020. Accessed December 2020. ftp://ftp.cmdl.noaa.gov/data/trace gases/co2/in-
situ/surface.

NOAA (National Oceanic and Atmospheric Administration). 2020c. Monthly mean CH4 concentrations for
Mauna Loa, Hawaii. Updated July 24, 2020. Accessed December 2020.

ftp://ftp.cmdl.noaa.gov/data/trace gases/ch4/flask/surface/ch4 mlo surface-flask 1 ccgg month.txt.

NOAA (National Oceanic and Atmospheric Administration). 2020d. Monthly mean N20 concentrations
for Barrow, Alaska; Mauna Loa, Hawaii; and the South Pole. Accessed December 2020.
www.esrl.noaa.gov/gmd/hats/insitu/cats/cats conc.html.

Rigby, M. 2017 update to data originally published in: Arnold, T., C.M. Harth, J. Miihle, A.J. Manning, P.K.
Salameh, J. Kim, D.J. Ivy, L.P. Steele, V.V. Petrenko, J.P. Severinghaus, D. Baggenstos, and R.F. Weiss.
2013. Nitrogen trifluoride global emissions estimated from updated atmospheric measurements. P. Natl.
Acad. Sci. USA 110(6):2029-2034. Data updated December 2017.

Schilt, A., M. Baumgartner, T. Blunier, J. Schwander, R. Spahni, H. Fischer, and T.F. Stocker. 2010. Glacial-
interglacial and millennial scale variations in the atmospheric nitrous oxide concentration during the last
800,000 years. Quaternary Sci. Rev. 29:182-192. www.ncdc.noaa.gov/paleo-search/studv/8615.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

16

-------
Steele, L.P., P.B. Krummel, and R.L. Langenfelds. 2002. Atmospheric methane record from Shetland
Islands, Scotland (October 2002 version). In: Trends: A compendium of data on global change. Oak
Ridge, TN: U.S. Department of Energy. Accessed September 13, 2005. http://cdiac.ess-
dive.lbl.gov/trends/atm meth/csiro/csiro-shetlandch4.html.

Steele, L.P., P.B. Krummel, and R.L. Langenfelds. 2007. Atmospheric C02 concentrations (ppmv) derived
from flask air samples collected at Cape Grim, Australia, and Shetland Islands, Scotland. Commonwealth
Scientific and Industrial Research Organisation. Accessed January 20, 2009. http://cdiac.ess-
dive.lbl.gov/ftp/trends/co2/csiro.

Weiss, R.F., J. Muhle, P.K. Salameh, and C.M. Harth. 2008. Nitrogen trifluoride in the global atmosphere.
Geophys. Res. Lett. 35:L20821.

Ziemke, J.R., S. Chandra, B.N. Duncan, L. Froidevaux, P.K. Bhartia, P.F. Levelt, and J.W. Waters. 2006.
Tropospheric ozone determined from Aura OMI and MLS: Evaluation of measurements and comparison
with the Global Modeling Initiative's Chemical Transport Model. J. Geophys. Res. 111:D19303.

Technical Documentation: Atmospheric Concentrations of Greenhouse Gases

17

-------