The Keeling Curve and Global CO2

This post is taken from a presentation which I gave at a local geologic conference in 2104, with minor modifications.

Major findings of the CO₂ study:

• Atmospheric CO₂ is rising globally as a result of human activities, principally the burning of fossil fuels. 
• Atmospheric CO₂ concentration is now about 40% higher than pre-industrial CO2 concentration.
• CO₂ emissions from fossil fuels mostly originate in the Northern Hemisphere.  A global system of monitoring stations shows the dispersion of these emissions from the Northern Hemisphere to the Southern Hemisphere.
• Atmospheric CO₂ shows a seasonal cycle that is dominated by plant growth in the Northern Hemisphere.
• The amplitude of the seasonal cycle is increasing due to human agriculture.  Agriculture accounts for about 1/3 of the seasonal cycle.
• The Southern Hemisphere has a seasonal cycle that is the opposite polarity of the Northern Hemisphere, but is very weak due to a smaller land mass and less agriculture.
• Atmospheric CO₂ concentrations and carbon isotope ratios are changing much slower than expected if all human carbon emissions remained in the atmosphere.  About half of human carbon emissions are being absorbed by carbon reservoirs (oceans, soils and biomass).  Carbon isotopes show that large volumes of atmospheric carbon are freely exchanged with carbon in carbon reservoirs.  
• The size of the carbon reservoirs exchanging carbon with the atmosphere can be estimated through the dilution of human-derived carbon isotopes in the atmosphere.  The calculation indicates that these carbon reservoirs contain about 7 times the quantity of carbon in the atmosphere.  This solution is about 40% larger than estimates using other methods.
• Human carbon emissions will continue.  Forecasts indicate that atmospheric CO₂ will reach 450 ppm by the year 2036.
• After filtering the seasonal cycle from the carbon isotope data, a multi-year cycle remains.  The multi-year cycle can be correlated with the El Nino climate cycle.  The El Nino cycle changes the rate at which atmospheric carbon is absorbed by the Pacific Ocean, and changes isotopic composition of the atmosphere.

The Keeling Curve and Global CO2

S. D. Robbins                       May 15, 2014

Abstract

The Keeling Curve is a remarkable series of atmospheric CO₂ measurements taken at Mauna Loa, Hawaii, from 1960 to the present.  The curve shows seasonal cycles and a steady rise in the concentration of CO₂, beginning about 315 ppm and currently approaching 400 ppm.   Long-term CO₂ records are also available from a number of other observatories, located from the Arctic Ocean to the South Pole.   Integration of the global dataset with carbon emissions data provides additional insights about the world’s carbon cycle. 

Atmospheric CO₂ concentrations and CO₂ carbon isotopes show seasonal and long term trends which vary by latitude.  The seasonal cycle is strongest in the Northern Hemisphere, and the Northern Hemisphere leads the Southern Hemisphere in terms of rising CO₂.  People and plants in the Northern Hemisphere cause changes in atmospheric CO₂ which propagate from the Northern Hemisphere to the Southern Hemisphere. The dispersion of CO₂ from human sources can be seen as a progression through the global data.  This progressive change is seen in bulk concentration of CO₂, in carbon isotopes of CO₂, and the amplitude of the seasonal cycle.  Long-term changes in global CO₂ are consistent with known volumes of fossil fuel emissions.   A simple model can be constructed based solely on human carbon emissions and agricultural biomass, that matches the observed seasonal cycle and long-term trends in bulk CO₂,.  This model shows that it is reasonable to conclude that human activities are influencing carbon dioxide in the atmosphere.  The Energy Information Agency (EIA) predicts rising carbon emissions for the foreseeable future, from about 35 gigatonnes CO₂ annually to nearly 50 gigatonnes CO₂ by the year 2040, in the base-case forecast.

Carbon dioxide from fossil fuels and deforestation carries a distinctive isotopic signature, which marks the movement of man-made CO₂ through the atmosphere and carbon reservoirs (soils, biomass, and oceans).  This movement of carbon, as seen in both carbon isotope data and bulk CO₂ data, reveals complexity in the carbon cycle.  Discrepancies between the datasets imply the active exchange of carbon between the atmosphere and carbon reservoirs.  More than 85% of anthropogenic CO₂ emissions, as tagged by carbon isotopes, do not remain in the atmosphere, but are absorbed by carbon reservoirs.  However, some of the anthropogenic carbon in the atmosphere is exchanged for natural carbon from carbon reservoirs, so that atmospheric CO₂ concentration is maintained at a level equivalent to about 44% of cumulative annual CO₂ emissions over the long term.  The size of carbon reservoirs is estimated at more than 7 times the volume of carbon present in the atmosphere, based on a dilution calculation of anthropogenic carbon isotopes in the atmosphere.  The role of the ocean in exchanging carbon with the atmosphere is illustrated by the correlation of atmospheric carbon isotope data with the Oceanic Niño Index (ONI), which is a measure of the El Niño/La Niña climate cycle based on sea-surface temperatures.

 

Understanding the patterns of atmospheric CO₂ may provide a tool for recognizing and measuring changes in global climate.  Additional monitoring of carbon reservoirs, particularly of the world’s oceans, will be necessary to develop a comprehensive model of the earth’s carbon cycle. 

 -------------------------------------------------------------------------------------------------------------------------

A Few Words about CO₂ Carbon Isotopes

There are two stable isotopes of carbon, C13 and C12.   C12 is the more abundant isotope; the natural ratio of C12 to C13 is about 99 to 1.  The standard measure of carbon isotopes compares the C12/C13 isotope ratio of the sample in question to the C13/C12 ratio of a standard limestone, according to the expression:

d C13/C12 = ((C13/C12 sample  / C13/C12 standard) – 1)*1000.

This expression, commonly termed “del 13”, amplifies small but meaningful differences in the isotopes, which are diagnostic of certain processes and occurrences of carbon.  The standard is a uniform Cretaceous limestone with a d 13 value defined as zero.   Positive values indicate a heavier composition, i.e., a greater concentration of C13 than the standard.  Negative values indicate a lighter composition, i.e., a smaller concentration of C13 than the standard.

Plants fractionate carbon, favoring the lighter isotope C12.  Anything derived from plants, including oil, gas, and coal (and algae, animals and people) carries a light (negative) d C13/C12 signature.  Limestone carries a d  C13/C12 ratio near zero.  The atmosphere, in 1977, had a d C13/C12 ratio of about -7.5; it is currently about -8.3, reflecting the influence of fossil fuels.  Oceans have a slightly positive d C13/C12 ratio of dissolved inorganic carbon, although Northern Hemisphere waters show a negative ratio due to the greater use of fossil fuels in the Northern Hemisphere.  Fossil fuel CO₂ emissions and CO₂ emissions from deforestation carry a very light d C13/C12, in the range of -25 to -28.  The distinctive isotopic signature of CO₂ from fossil fuels and deforestation is useful in tracking the movement of carbon through the atmosphere and oceans.  Boden, Marland and Andres (2013) published estimates of the annual CO₂ released by fossil fuels and the d C13/C12 ratio of those emissions.  Those estimates were used in this work.  

Part 1:  The Global Record

Global CO₂ is rising, and the isotopic composition of atmospheric CO₂ is becoming lighter.

A network of observatories, mostly operated by Scripps Oceanographic Institute, monitors global atmospheric CO₂ concentrations and CO₂ carbon isotopes.   Bulk CO₂ has been monitored since 1957, and CO₂ carbon isotopes since 1977.  In all figures, data from the Northern Hemisphere is indicated in cool colors, and data from the Southern Hemisphere is indicated with warm colors.

Global CO₂ observations show a seasonal cycle and a steady rise in the concentration of CO₂.  Today’s concentration of atmospheric CO₂ is about 25% higher than in 1957, and about 40% higher than in 1800. 

The carbon isotope ratio (d C13/C12) of atmospheric CO₂ is becoming lighter, which is consistent with the isotopic signature of fossil fuels mixing with the atmosphere.

 The amplitude of the CO₂ seasonal cycle is largest in the high latitudes of the Northern Hemisphere, and diminishes southward.  The polarity of the Northern Hemisphere cycle persists to about 30 degrees South Latitude.   From that point southward, the polarity of the cycle is reversed, but with low amplitude.

Part 2:  Long Term Trends

The Northern Hemisphere leads the Southern Hemisphere in rising CO2 values and falling CO2 carbon isotope values.

The Northern Hemisphere holds 2/3 of the world’s landmass, and nearly 90% of the world’s population, fossil-fuel consumption, and agriculture.  People and plants in the Northern Hemisphere cause changes in atmospheric CO₂ which propagate from the Northern Hemisphere to the Southern Hemisphere.

The seasonal cycle in bulk CO₂ was removed by taking a one-year rolling average at each observatory.  The Northern Hemisphere leads the Southern Hemisphere in rising CO₂.  The concentration of CO₂ is highest in the Arctic, and is progressively lower by latitude to the South Pole.  The progression marks the dispersion of fossil fuel emissions from the Northern Hemisphere to the Southern Hemisphere.

The seasonal cycle in the CO₂ carbon isotope ratio was removed by taking a one-year rolling average at each observatory.   The Northern Hemisphere leads the Southern Hemisphere in terms of falling d C13/C12 of CO₂ (becoming isotopically lighter).  The isotopic composition of CO₂ is lightest near the North Pole, and becomes progressively heavier to Antarctica.  The progression marks the dispersion of fossil fuel emissions from the Northern Hemisphere to the Southern Hemisphere.

A simple model was constructed to investigate the plausibility of the idea that human activity causes changes in atmospheric CO₂.  The model begins at the global baseline CO₂ concentration in 1970.  Model inputs included 60% of global fossil fuel emissions, allocated to Northern and Southern Hemispheres by GDP.  Global agricultural biomass was scaled by year to human population, and allocated to the Northern and Southern Hemispheres by population.  Volumes of carbon were converted to CO₂ concentration by hemisphere, and defined the concentration at the poles.  Concentrations at intermediate latitudes were created by mixing concentrations from each hemisphere, with weighting by latitude.  The ease with which the model was created suggests that human activity is plausibly responsible for much of the change in atmospheric CO₂.

Part 3:  Seasonal Cycle

The Northern Hemisphere dominates the global CO₂ seasonal cycle.

The Keeling Curve is characterized by a strong seasonal cycle, dominated by the Northern Hemisphere.  The concentration of atmospheric CO₂ falls during the Northern Hemisphere growing season, when land plants remove carbon from the air through photosynthesis.  The concentration of CO₂ rises in the fall, winter and spring as decay returns carbon to the atmosphere as CO₂.

The CO₂ carbon isotope ratio d C13/C12 shows a strong seasonal cycle as a mirror image of the cycle in bulk CO₂ concentration.  Land plants in the Northern Hemisphere strongly fractionate carbon isotopes.  Plants preferentially absorb C12 during the growing season, raising the d C13/C12 ratio of the atmosphere.  During decay, plants return C12 to the atmosphere, and atmospheric d C13/C12 falls.

 

Amplitude of the seasonal cycle is relatively small in the Southern Hemisphere, reflecting a smaller land mass and sparse population.  Seasonal cycles in low latitudes of the Southern Hemisphere follow the polarity of the Northern Hemisphere, but with a phase shift.  Seasonal cycles in high latitudes of the Southern Hemisphere carry the opposite polarity to the Northern Hemisphere.

Amplitude of the seasonal cycle is increasing over the past 40 years, particularly at high northern latitudes.  The increase in amplitude correlates well to the increase in human population over the past 40 years.   By inference, the increase in seasonal amplitude also correlates to a proportional increase in human agriculture.  The correlation implies that agriculture accounts for about one-third of the amplitude of the seasonal CO₂ cycle. 

The polarity reversal of the seasonal cycle occurs at about 30 degrees South Latitude, near the southern boundary of the Hadley convection cell.  The atmosphere north of -30 degrees latitude contains air which is mixed with air from the Northern Hemisphere; CO₂ concentrations and carbon isotopes follow the seasonal cycle of the Northern Hemisphere.  The atmosphere south of -30 degrees latitude carries the seasonal cycle of the Southern Hemisphere.  This finding suggests that additional CO₂ observatories between -30 degrees and -40 degrees south latitude could monitor climate-change induced expansion of Hadley circulation, by detecting air from the Northern Hemisphere, according to the Northern Hemisphere seasonal CO₂ cycle.

Part 4:  CO₂ Emissions

Long-term changes in atmospheric CO₂ are consistent with known volumes of human CO₂ emissions.

The rate of human CO₂ emissions is increasing.  Anthropogenic CO₂ emissions, including deforestation, have grown from about 5 gigatonnes annually in 1900 to about 38 gigatonnes in 2009.  The greatest part of that increase occurred in the last 50 years.  

The average CO2 concentration of the Northern Hemisphere leads the Southern Hemisphere by 2.5 to 4 ppm CO₂.  Net annual fossil fuel emissions in the Northern Hemisphere, converted to CO₂ concentration, neatly match the difference in CO₂ concentration between the hemispheres.  Deforestation, (which is more prevalent in the Southern Hemisphere) was not included in the emissions numbers, which might account for the growing discrepancy in recent years.

Despite international efforts to reduce carbon emissions, global industrial CO₂ emissions are rising sharply.  According to the EIA base economic forecast, fossil fuel CO₂ emissions are expected to rise 45% by the year 2040, from 33 gigatonnes to 48 gigatonnes per year.  Emissions including deforestation bring the total in 2040 to 53 gigatonnes, assuming a constant rate of deforestation from 2005 to 2040.

Global average CO₂ is rising at a rate equal to about 44 percent of annual CO₂ emissions, including deforestation.  The forecast of future CO₂ concentrations, based on expected emissions, calls for world average CO₂ to exceed 450 ppm around the year 2036.  

Part 5: The Carbonsphere

The Carbonsphere consists of atmospheric carbon and all reservoirs (ocean, biomass, and soils) freely exchanging carbon with the atmosphere.

The global mix of fossil fuels has a d C13/C12 value of about -28.  Deforestation is assumed to have a d C13/C12 value of about -25.  These contrast sharply with the atmospheric d C13/C12 value of about -8, and slightly positive oceanic d C13/C12 values.  The distinctive isotopic signature of human carbon emissions allows us to track the movement of carbon through the atmosphere, and to detect the exchange of carbon with carbon reservoirs on the earth’s surface.

A 2-year lag is required for the concentration of bulk CO₂ to equilibrate from sources in the Northern Hemisphere to the Antarctic.

Differences in the behavior of bulk CO₂ and CO₂ carbon isotopes indicate the active role of carbon reservoirs in the ocean, plants, and soil in exchanging carbon with the atmosphere. 

Bulk carbon requires about a 2-year lag for CO₂ concentration to equilibrate from the Arctic to the Antarctic.  In contrast, CO₂ carbon isotopes require an 8-year lag for equilibration. 

The difference indicates that the specific molecules released by fossil fuels are cycled through carbon reservoirs and replaced in the atmosphere by other molecules from those reservoirs.  The difference in equilibration lag of bulk carbon and carbon isotopes indicates residency time in those reservoirs.

Atmospheric CO₂ concentration rises at a rate equal to about 44% of human CO₂ emissions, including deforestation.  If all human carbon emissions remained in the atmosphere, the concentration of atmospheric CO₂ would be much higher.  

Atmospheric CO₂ d C13/C12 falls at a rate incorporating about 12% of human CO₂ emissions. 

If all human carbon emissions remained in the atmosphere, the concentration of atmospheric CO₂ would be much higher, and the d C13/C12 isotope ratio would be much lower.  Calculations indicate that atmospheric CO₂ rises at a rate equal to about 44% of human CO2 emissions.  In contrast, CO₂ carbon isotopes indicate that only 12% of human carbon emissions remain in the atmosphere.  More than 85% of human carbon emissions, as tagged by carbon isotopes do not remain in the atmosphere, but are cycled into other carbon reservoirs.  At the same time, natural carbon, carrying a heavier isotopic signature, is exchanged from the carbon reservoirs, maintaining a bulk CO₂ concentration accumulating at a rate of 44% of human CO₂ emissions.

In the inverse of the calculation above, the d C13/C12 isotope ratio of the atmosphere shows the total volume of carbon reservoirs interacting with the atmosphere.  The calculation determines the total reservoir volume necessary to produce the observed dilution of d C13/C12 from human emissions in the air.  Carbonsphere reservoirs are assumed to be in equilibrium with the atmosphere, which is demonstrated by the relatively good fit to the solution for 30 years.  The model assumes a balance of fractionation between the carbonsphere reservoirs and the atmosphere.  The solution calls for a 6000 gigatonne carbonsphere in 1979 (including the atmosphere).  This solution is about 40% larger than estimates based on carbon inventory methods.

Part 6:  Finding Niño

The CO₂ carbon isotope record can be correlated with the El Niño/La Niña climate cycle, indicating large volumes of carbon exchange between the atmosphere and the tropical Pacific Ocean.

Multi-annual cycles, or “waves” are present in the CO₂ carbon isotope data after removing the seasonal cycle.  Lower amplitude but correlative waves also exist in the bulk CO₂ chart.  Periods of rapidly falling d C13/C12 correspond to El Niño climatic events, suggesting variability in the rate at which the tropical Pacific Ocean takes up or releases C12 to the atmosphere.  The amplitude of the waves interrupts and sometimes reverses the secular trend, indicating that the volumes of carbon exchanged sometimes exceed the volume of human carbon emissions.

A series of mathematical operations can reduce the global CO₂ isotope record to a single trace which indicates the rate at which atmospheric carbon is exchanged with the Carbonsphere.  Assuming no fractionation in the exchange process, positive values indicate a faster rate of absorption of C12 by the Carbonsphere.  Negative values indicate a slower rate of C12 absorption by the Carbonsphere.  With a slight time lag, the trace can be correlated with the Oceanic Niño Index (ONI).  The ONI is a measure of sea surface temperatures published by NOAA, indicating the prevalence of El Niño or La Niño conditions. 

The trace derived from the d C13/C12 isotope data indicates variability in the rate of exchange of C12 between the atmosphere and carbon reservoirs. That curve correlates well with the Oceanic Niño Index, a measure of the El Niño/El Niña climate cycle.  La Niña conditions, which are characterized by abnormally cool sea surface temperatures, correspond with accelerated absorption of atmospheric C12 by the ocean.  El Niño conditions, characterized by warm sea surface temperatures, correspond to decreased absorption of atmospheric C12 by the ocean. 

The quality and quantity of oceanic carbon data is weak in comparison to atmospheric CO₂ data.  Oceanic carbon data is limited by a lack of continuous readings or fixed observation sites, and is strongly influenced by local biological activity.  Nevertheless, at the broadest scale, oceanic dissolved inorganic carbon is isotopically lighter in the Northern Hemisphere, reflecting the influence of fossil fuels in the Northern Hemisphere.  Improved, systematic data collection in the oceans and other carbon reservoirs will be necessary to develop a comprehensive model of the earth’s carbon cycle.

Note: Since the original poster was presented in May, 2014, I have noticed that the low d C13/C12 values are located in the Atlantic Ocean, and may be associated with the Greenland Current.  It is possible that these values reflect a natural process, such as a significant volume of glacial meltwater lowering the d C13/C12 value of the seawater.

References:

Andres, R.J., T.A. Boden, and G. Marland. 2012.  Monthly

Fossil-Fuel CO2 Emissions: Mass of Emissions Gridded by One Degree

Latitude by One Degree Longitude.  Carbon Dioxide Information Analysis

Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak

Ridge, Tenn., U.S.A.  doi: 10.3334/CDIAC/ffe.MonthlyIsomass.2012

ftp://cdiac.ornl.gov/pub/fossil_fuel_CO2_emissions_gridded_monthly_del13C_v2012/

Andres, R.J., T.A. Boden, and G. Marland. 2009.  Monthly

Fossil-Fuel CO2 Emissions: Mass of Emissions Gridded by One Degree

Latitude by One Degree Longitude.  Carbon Dioxide Information Analysis

Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak

Ridge, Tenn., U.S.A.  doi: 10.3334/CDIAC/ffe.MonthlyIsomass.2009

ftp://cdiac.ornl.gov/pub/fossil_fuel_CO2_emissions_gridded_monthly_del13C_v2009

Andres, R.J., T.A. Boden, and G. Marland. 2012.  Annual Fossil-Fuel CO2 Emissions: Global Stable Carbon Isotopic Signature.  Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A.

doi: 10.3334/CDIAC/ffe.db1013.2012

http://cdiac.ornl.gov/ndps/db1013_v2012.html

Boden, T.A., Andres, R.J., and G. Marland 2013. Global CO2 Emissions from Fossil-Fuel Burning, Cement Manufacture, and Gas Flaring: 1751-2010. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A.   

http://cdiac.ornl.gov/ftp/ndp030/global.1751_2010.ems

Boden, T.A., G. Marland, and R.J. Andres. 2013. Global, Regional, and National Fossil-Fuel CO2 Emissions. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A. doi 10.3334/CDIAC/00001_V2013

http://cdiac.ornl.gov/trends/emis/overview_2010.html

Ciattaglia, L., C. Rafanelli, H. Rodriguez, and J. Araujo. 2010. Atmospheric CO2 record from continuous measurements at Jubany Station, Antarctica, in Trends Online: A Compendium of Data on Global Change, Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A.

http://cdiac.esd.ornl.gov/trends/co2/jubany.html

Climate Prediction Center Internet Team, 2013, ONI Index, NOAA/National Weather Service, NOAA Center for Weather and Climate Prediction , Climate Prediction Center, 5830 University Research Court, College Park, Maryland 20740.

http://www.cpc.ncep.noaa.gov/data/indices/oni.ascii.txt

http://www.nwfsc.noaa.gov/research/divisions/fe/estuarine/oeip/cb-mei.cfm

Colombo, T., and R. Santaguida. 1998. Atmospheric CO2 record from in situ measurements at Mt. Cimone. In Trends: A Compendium of Data on Global Change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A.

http://cdiac.esd.ornl.gov/trends/co2/mtcim.html

Conti, J., et al, 2013.  International Energy Outlook, U. S. Energy Information Administration, Office of Energy Analysis, U.S. Department of Energy, Washington, D.C.    DOE/EIA-0484(2013)

http://www.eia.gov/forecasts/ieo/

http://www.eia.gov/forecasts/ieo/pdf/0484(2013).pdf

Gaudry, A., V. Kazan, and P. Monfray. 1996. Atmospheric CO2 record from in situ measurements at Amsterdam Island. In Trends: A Compendium of Data on Global Change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A.

http://cdiac.esd.ornl.gov/trends/co2/amster.html

Houghton, R.A. 2008. Carbon Flux to the Atmosphere from Land-Use Changes: 1850-2005. In TRENDS: A Compendium of Data on Global Change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A.

http://cdiac.ornl.gov/trends/landuse/houghton/houghton.html

Keeling, C.D., S.C. Piper, R.B. Bacastow, M.  Wahlen, T.P. Whorf, M. Heimann, and H. A. Meijer, Exchanges of atmospheric CO2 and 13CO2 with the terrestrial biosphere and oceans from 1978 to 2000.

I. Global aspects, SIO Reference Series, No. 01-06, Scripps Institution of Oceanography, San Diego, 88 pages, 2001.

http://scrippsco2.ucsd.edu/data/atmospheric_co2.html

Keeling, R.F., S.C. Piper, A.F. Bollenbacher and J.S. Walker. 2008. Atmospheric CO2 records from sites in the SIO air sampling network. In Trends: A Compendium of Data on Global Change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A.

http://cdiac.esd.ornl.gov/trends/co2/sio-keel.html

Sieminski, A., 2013, International Energy Outlook 2013, Center for Strategic and International Studies, Washington, DC.

http://www.eia.gov/pressroom/presentations/sieminski_07252013.pdf

U.S. Energy Information Agency, Data Tables, U.S. Energy Information Agency, Office of Energy Analysis, U.S. Department of Energy, Washington D.C. data tables, 2014

http://www.eia.gov/oiaf/aeo/tablebrowser/