Determination of the distribution of CO2, CH4, and other greenhouse gases in the atmosphere

Knowledge of today’s carbon sources and sinks, their spatial distribution and their variability in time is one of the essential ingredients for predicting the future carbon dioxide (CO2) atmospheric concentration levels, and in turn radiative forcing of climate change by CO2. The distribution of atmospheric CO2 reflects both spatial and temporal evolutions as well as the magnitude of surface fluxes (Tans et al., 1990). The traditional atmospheric top-down approach (e.g. Gurney et al., 2004) uses atmospheric transport models to determine the spatio-temporal land-air surface flux distribution that gives the best match to a global set of atmospheric CO2 data. In principle, it is thus possible to estimate these fluxes from atmospheric CO2 concentration, provided that atmospheric transport can be accurately modeled. However, this approach is currently limited by the sparse and uneven distribution of the global flask sampling programs. Moreover, it requires an accurate modeling of atmospheric transport, which still suffers from limitation, especially along the vertical. Densely sampling the atmosphere in time and space, satellite measurements of the distribution of global atmospheric CO2 concentration could in principle provide a way to address both issues.

Since 2001, the ARA team has pioneered a research on retrieving global scale atmospheric CO2 concentration from infrared space observation, first in terms of mid-to-upper tropospheric content using the TOVS, AIRS (2003-2007) and IASI (since July 2007) instruments and then, more recently, in terms of vertical profiles of CO2 using the limb-viewing ACE-FTS instrument. Some of the retrieval algorithms are now being extended to the retrieval of CH4. The team also participates in the design of a CO2 lidar mission (project A-SCOPE) and of a CH4 lidar mission (project MERLIN) by studying the conditions of an optimal coupling of an infrared sounder and the lidar, as well as to the exploitation of the TANSO instrument launched onboard JAXA/ GOSAT on 23 January 2009, which is the first space mission specifically designed for CO2 monitoring.

Mid-tropospheric integrated column from TOVS, AIRS and IASI sounders

Although infrared sounders have historically been designed for meteorological soundings (temperature, water vapor, ozone), the impact of CO2 variations can be seen on some satellite records, in particular temperature estimated from infrared observations made in CO2 absorption bands. The main difficulty in estimating global distribution of CO2 from infrared sounders comes from the fact that CO2 infrared measurements are sensitive to both temperature and CO2 variations. Independent information on temperature is thus needed to allow separating these two effects. As a consequence, higher-quality retrievals are expected in the tropical band because of the low variability of the temperature profiles band compared to the extratropics.

The first interpretation of infrared radiances in terms of CO2 concentration were performed by Chédin et al. (2002), yielding the first observation of the year-to-year increase of CO2 from space. Chédin et al. (2002) used observations from the first generation NOAA10/TOVS instruments to analyze the differences between the satellite observations and simulations from the 4A radiative transfer model using collocated radiosonde data and fixed gas concentration as the prime input to derive seasonal cycles and trends of CO2, N2O and CO in three latitudinal bands. Chédin et al. (2003) then used a non linear inference scheme based on neural network to differentiate between temperature and CO2 by including Microwave Sounding Unit (MSU) radiances in the process, since microwave radiance are insensitive to CO2 but not to temperature. A key point of this approach is that no use is made of prior information in terms of CO2 seasonality, trend, or geographical patterns. Four years of mid-to-upper tropospheric CO2 content covering the 1987-1991 time frame between 20S and 20N were retrieved showing good agreement with existing in situ (aircraft) observations: phase and amplitude of the seasonal variations, impact of El Niño Southern Oscillation (ENSO) events, etc... A mean CO2 growth rates of 1.75 ppm/year was also retrieved, in good agreement with in situ measurements. Accuracy of the retrievals from this first generation of infrared sounders, around 1%, was however limited by the low spectral resolution of the TOVS instruments, and by the lack of channels only sensitive to CO2.

Time variations of monthly CO2 from TOVS
Time variations of monthly CO2 concentration (crosses) in the tropics retrieved from TOVS (Chédin et al., 2003) for: (a) the northern tropics (0°-20°N); (b) the southern tropics (0°-20°S). The dashed line shows the long-term trend and the solid line shows the sum of the trend and of the harmonics.

The launch of AIRS and AMSU onboard NASA/Aqua satellite in May 2002 brought a new step in the monitoring of trace gases from space, and several studies presented CO2 estimates from AIRS. Crevoisier et al. (2004) extended the stand-alone approach developed by Chédin et al. (2003) to derive a mid-to-upper tropospheric content of CO2, using the Aqua/AMSU microwave observations to bring the necessary information on atmospheric temperature. The measurement standard deviation was about 2.5 ppm on a monthly time scale.

Monthly variation of mid-tropospheric CO2 and CH4
Monthly variation of mid-tropospheric CO2 (top) and CH4 (bottom) as derived from the AIRS (orange) and IASI (red) instruments. The trend and seasonal cycle of both gases are well seen.

With its very high spectral resolution, IASI, launched onboard the European MetOp satellite in October 2006 and declared operational in July 2007, provides fourteen channels in the 15μm band highly sensitive to CO2 with reduced sensitivities to other atmospheric variables. IASI observations, sensitive to both CO2 and temperature, are used in conjunction with observations from AMSU, also flying onboard MetOp and only sensitive to temperature, to decorrelate both signals. The precision of the retrieval is estimated to be about 2.0 ppmv (0.5%) for a 5°×5° spatial resolution on a monthly time scale (Crevoisier et al. 2009a). IASI giving the opportunity to use several channels specifically sensitive to methane (CH4), the retrieval procedure as been adapted to the study of CH4 distribution (Crevoisier et al. 2009b), with a monthly precision of about 16 ppbv (~1 %).

Show/Hide Temporal CH4 slideshow

Options availables
Use keyboard to control the gallery : ⇐ Previous photo, ⇒ Next photo

Hide slideshow
Correlation between CO2 and CH4
Correlation between CO2 and CH4 as retrieved from IASI in winter (left) and summer (right).

Features of the retrieved CO2-CH4 space-time distributions include: (1) a CO2 trend of ~2.1 ppmv.yr-1 in average, and a CH4 trend of ~10 ppbv.yr-1 in the last couple of years, which confirms the recent increase of methane detected at surface stations; (2) a strong seasonal cycle in the northern tropics, and a lower seasonal cycle in the southern tropics, in agreement with in-situ measurements, in particular those performed at 11 km during the JAL/CONTRAIL programs (Matsueda et al., 2008; Machida et al., 2008; Sawa et al. 2008) ; more specifically, a comparison between AIRS and IASI retrievals highlights the time-lag of CO2 cycle while transported from the surface to the upper troposphere; (3) a latitudinal decrease from 20N to 20S lower than what is observed at the surface but in excellent agreement with tropospheric aircraft measurements; (4) geographical patterns in good agreement with simulations from atmospheric transport and chemistry models (MOZARTv2, TM5), but with a higher variability; (5) signatures of CO2 and CH4 emissions transported to the troposphere such as CO2 emissions from biomass burning, or a large plume of elevated tropospheric methane south of the Asian continent, which might be due to Asian emissions from rice paddies uplifted by deep convection during the monsoon period and then transported towards Indonesia. Moreover, these retrievals, performed from the same instrument and with the same retrieval process, provide the means to study the correlation between CO2 and CH4, in particular its seasonal variation over regions of specific interest, which leads the way to a multi-species study of surface fluxes and atmospheric transport.

In addition to bringing an improved view of CO2 and CH4 distribution, these results from IASI should provide an additional means to observe and understand atmospheric transport pathways of CO2 from the surface to the upper troposphere.

Vertical profile from ACE-FTS

Observation of the CO2 concentration has recently been extended to that of its vertical distribution by using the limb viewing instrument Atmospheric Chemistry Experiment - Fourrier Transform Interferometer (ACE-FTS),working in solar occultation mode, and flying on board the Canadian satellite SCISAT to retrieve CO2 vertical profiles from about 5 km to 25 km.

Major limitations of our present knowledge of the global distribution of carbon dioxide (CO2) in the atmosphere are the uncertainty in atmospheric transport mixing and the sparseness of in situ concentration measurements. Atmospheric CO2 is a good air transport tracer, therefore a precise determination of its vertical distribution has two interests: extending CO2 profiles measurement to the global scale and improving large scale atmospheric transport knowledge. Limb viewing space borne sounders, observing the atmosphere along tangential optical paths, offer a vertical resolution of a few kilometers for profiles, which is much better than currently flying or planned nadir sounding instruments can achieve.

The Atmospheric Chemistry Experiment Fourier Transform Spectrometer (ACE-FTS, launched in August 2003) provides solar occultation measurements with a good vertical resolution and records atmospheric spectra over a wide spectral domain from 700 cm-1 to 4400 cm-1 with a 0.02 cm-1 spectral resolution and a high signal to noise ratio. In order to interpret these measurements in terms of CO2 vertical profiles, two main difficulties must be overcome: (i) the accurate determination of the instrument pointing parameters (tangent heights) and pressure/ temperature profiles independently from an a priori CO2 profile, and (ii) the potential impact of uncertainties in the temperature knowledge on the retrieved CO2 profile. The first difficulty has been solved using, for the first time, the N2 collision induced continuum absorption near 4 µm to determine tangent heights, pressure and temperature from the ACE-FTS spectra. The second difficulty has been solved by a careful selection of CO2 spectral micro-windows.

The retrieval method is based on optimal estimation using 4A/OP-limb radiative transfer model for direct and inverse modelling. First CO2 vertical profiles have recently been retrieved from ACE-FTS measurements in the altitude range 5-25 km. Averaged over a month for 10° latitude bands, they show a theoretical 2 ppm relative precision and a vertical resolution around 2 km. Comparisons with collocated aircraft measurements (CARIBIC, Brenninkmeijer et al., 2007) and CO2 profiles from FLEXPART air transport model (Stohl et al., 2005) confirm these results (see P.-Y. Foucher, 2009 and PhD thesis for more details). These new developments open several research ways: creation of a 10-year CO2 vertical profiles data base with an optimal spatial and vertical resolution; improvement of air transport model simulations through systematic comparisons with observations; association with CO2 nadir measurements (GOSAT, OCO) to study CO2 lower troposphere distribution; study of local phenomenon as biomass burning injection heights.

Temporal evolution of CO2 concentration
Temporal evolution of CO2 concentration in the high troposphere (8-9 km). Blue boxes: ACE-FTS results; red boxes: CARIBIC measurements. The solid and dashed lines result from a curve fitting by a parametric periodic function.
Comparison between CO2 vertical profiles retrieved
Comparison between CO2 vertical profiles retrieved from ACE-FTS (red) and simulated by the Flexpart model (blue) in spring. The best agreement is obtained when applying a one month lag to the Flexpart simulations which could denote a problem in the modelled intensity of the vertical transport.

Impact of biomass fires and their impact on the carbon cycle

Vegetation fires (agriculture, deforestation) emit amounts of trace gases (CO2, CO, CH4) and aerosols of an order of magnitude comparable to that produced by the combustion of fossil fuels. They affect the whole climate by inducing retroactions still not well understood. The ARA team has been first to bring into evidence a “Daily Tropospheric Excess” (DTE) in the concentration of CO2 in the mid-troposphere over the tropical continents and to link it to fires owing to the analysis of the twice daily observations made by the NOAA/TOVS instruments (Chédin et al., 2005, 2008). A high correlation (R² = 0.8) has been found between the annual mean DTE and fire CO2 emissions from the Global Fire Emission Dataset (GFEDv2, van der Werf et al., 2006) at continental scale over most regions of the tropics leading to the idea that DTE data can be very useful as a quantitative proxy of fire emission spatial patterns.

Monthly Variations of CO2
Seasonal mean Daily Tropospheric Excess (DTE - difference between 19h30 and 07h30 local time) of mid tropospheric concentration of CO2 over the tropics due to fires (ppm) averaged over the period January 1987-December 1990. Spatial resolution: 5°×5°, 1° by 1° moving average.

Recently, the mid-troposphere diurnal anomaly observed from space between evening and morning CO2 columns has been investigated at regional scale for ten regions of southern Africa with the aim of analyzing its properties as a proxy of fire emission. A possible mechanism for CO2 concentration being higher in the evening than in the morning is that hot convective fire plumes inject emissions directly into the troposphere during the afternoon peak of fire activity, seen by the satellite in the evening, and then diluted by large-scale atmospheric transport before the next satellite pass in the morning. 3D simulations of the DTE signal by the LMDz General Circulation Model, in which a pyro-thermal plume model is activated (Rio et al., 2010), confirm the observations. A large fraction of fire products is directly injected in the mid-troposphere, well above the boundary layer. This rapid uplift of CO2, combined with atmospheric transport patterns in southern Africa during the dry season, characterized by a fluctuating continental gyre, produces a daily DTE signal mainly positive above the source region and either positive or negative outside of the source region. On a monthly mean, this results in a persistent DTE signal above the source region of an order of 1ppm, while the impact of large-scale advection vanishes.

Simulated integrated CO2 concentration
Simulated integrated CO2 concentration from biomass burning which would be retrieved from satellite at 7am (left), and 7pm (middle) and the difference between the concentration at 7pm and the one at 7am (right), on (from top to bottom) the 18th , 19th, 20th and 21th of July. Horizontal winds at 500 hPa averaged over night are drawn upon the modelled CO2 at 7am, and afternoon winds upon modelled CO2 at 7pm.

The DTE signal has then been compared with other datasets used to estimate burned areas and fire emissions (GFEDv2, the Global Fire Emission Database version 2, and L3JRC of the Joint Research Center) showing similar seasonal and annual patterns. DTE also displays regional scale interannual variability which correlates well with ENSO, as do fire emissions themselves. We conclude that the DTE signal might represent a quantitative proxy of fire emission spatial patterns, in particular before the ATSR or MODIS observation periods when better quality fire count and burned area data became available

GFEDv2 CO2 annual mean emissions
GFEDv2 CO2 annual mean emissions (in g CO2 m-2) averaged over the period 1997-2004 (van der Werf et al., 2006) versus annual mean DTE (in ppm) averaged over the period 1987-1990 for the 10 regions of the study and the 2 integrated ASn and ASs regions.


These research activities take place in the framework of the GMES inititiave and are, or have been, supported by various integrated European projects:

Monitoring Atmospheric Composition and Climate European Union

Links :

Last update : 2015/12/10