1 Interannual Variability of Mid-tropospheric CO2 from Atmospheric
2 Infrared Sounder
6 Xun Jiang1*, Moustafa T. Chahine2, Edward T. Olsen2, Luke L. Chen2,
7 and Yuk L. Yung3
20 1 Department of Earth & Atmospheric Sciences, University of Houston, USA
21 2 Science Division, Jet Propulsion Laboratory, California Institute of Technology, USA
22 3 Division of Geological and Planetary Sciences, California Institute of Technology,
23 Pasadena, USA.
26 * To whom all correspondence should be addressed. E-mail: firstname.lastname@example.org
28 To be Submitted to GRL, Nov 24 2009
3  Atmospheric Infrared Sounder (AIRS) offers a unique opportunity to investigate the
4 variability of mid-tropospheric CO2 over the entire globe. In this paper, we use AIRS data
5 to examine the interannual variability of CO2 and find significant correlations between
6 AIRS mid-tropospheric CO2 concentrations and large-scale atmospheric dynamics.
7 During El Niño events, mid-tropospheric CO2 over the central Pacific Ocean is enhanced
8 whereas it is reduced over the western Pacific Ocean as a result of the change in the
9 Walker circulation. The variation of AIRS CO2 in the high latitudes of northern
10 hemisphere is closely related to the strength of the northern hemispheric annular mode.
11 These results contribute to a better understanding of the influence of large-scale dynamics
12 on tracer distributions.
1 1. Introduction
3  The increasing level of atmospheric CO2 has a significant influence on global
4 climate change [Dickinson and Cicerone, 1986]. Observations show a global trend of
5 CO2 of approximately 2 ppm/year [Keeling et al., 1995]. Superimposed upon this trend
6 is an annual cycle resulting from the uptake and release of CO2 by vegetation whose
7 amplitude is greatest in the northern hemisphere (NH). In addition to the trend and annual
8 cycle, atmospheric CO2 also shows interannual variability. In this paper, we will
9 investigate the impact of the transport and large-scale dynamics on the interannual
10 variability of CO2.
12  El Niño and Southern Oscillation (ENSO) is the most important large-scale climate
13 interannual variability in the tropics. Atmospheric CO2 is influenced by the ENSO at the
14 surface [Bacastow 1976; Bacastow et al., 1980]. During El Niño (La Niña) events, the
15 atmospheric CO2 growth rate increases (decreases) at surface stations [Keeling et al.,
16 1995; Jones et al., 2001]. The reasons are as follows.
18  During El Niño events, ocean is a sink for CO2 [Feely et al., 1987; Inoue and
19 Sugimuray 1992; Wong et al., 1993; Feely et al., 1997; Feely et al., 2006]. Upwelling of
20 the cold nutrient-rich water off the South American coast ceases during El Niño, causing
21 the surface pCO2 concentration to decrease along the equator [Feely et al., 1987; 1997].
22 Meanwhile, the tropical land becomes dryer and warmer. As a result, the gross primary
23 productivity decreases and respiration increases over the land [Jones et al., 2001]. Thus
24 the terrestrial biosphere becomes a greater source of CO2 to the atmosphere during El
25 Niño [Keeling et al., 1995; Francey et al., 1995]. The net effect from land and ocean is
26 that the surface CO2 growth rate increases during El Niño events. Conversely, the surface
27 CO2 growth rate decreases during La Niña events.
29  The focus of these previous studies is on the influence of climate variability on the
30 CO2 concentration at a few stations near the surface. However, there is no previous
31 investigation on understanding the influence of El Niño on the mid-tropospheric CO2 on
1 a global scale. In this paper, we analyze the influence of ENSO on the mid-tropospheric
2 CO2 retrieved by the Atmospheric Infrared Sounder (AIRS). This global dataset provides
3 new insight on how ENSO influence the carbon interannual variability in the middle
4 atmosphere and contributes to our understanding of the vertical transport of tracers
5 between the surface and the mid-troposphere.
7  The global coverage of AIRS CO2 retrievals also allows us to investigate the
8 influence of large-scale dynamics on the CO2 concentration in the polar region. The polar
9 region has profound significance for climate. Current general circulation models (GCMs)
10 do not perform well in the polar region, especially in simulating the exchange between
11 the stratosphere and troposphere in that region [Meloen et al. 2003; Jiang et al., 2009].
12 Therefore, observations are essential for improving our understanding of the large-scale
13 dynamics in the polar region. However, most observations and measurements in the polar
14 region suffer from limited coverage both in time and space. The global distribution of
15 CO2 retrieved from AIRS offers a unique opportunity for studying the large-scale
16 dynamics in the polar region.
18  The annular modes are the most important source of interannual variability in the
19 polar region [Thompson and Wallace, 1998]. Jiang et al. [2008a, 2008b] applied
20 principal component analysis to the extratropical total column ozone from the combined
21 Merged Ozone Data product and the European Center for Medium-Range Weather
22 Forecasts assimilated ozone from Jan 1979 to Aug 2002. In both hemispheres, the first
23 two leading modes are nearly zonally symmetric and are related to the annular modes.
24 When the polar vortex is stronger (positive phase of the annular mode), there is less
25 ozone transported to the polar region due to a weaker Brew-Dobson circulation. Similar
26 results in extratropical column ozone are found in the Goddard Earth Observation System
27 Chemistry-Climate Model (GEOS-CCM) [Jiang et al., 2008a; 2008b]. Since there is no
28 chemical destruction of CO2, in contrast to O3, CO2 is a better tracer for investigating the
29 large-scale dynamics in the polar region. In this paper, we will examine the influence of
30 the northern hemispheric annular mode (see later discussion on its definition) on AIRS
31 mid-tropospheric CO2.
2 2. Data
3  AIRS is a cross-track scanning grating spectrometer with 2378 channels from 3.7 to
4 15.4 m with a 13.5 km field of view at nadir [Aumann et al., 2003]. Chahine et al.
5  found that the range 690-725 cm-1 is best for selecting the main channel set to
6 retrieve the mid-tropospheric CO2 mixing ratio. The mixing ratios of mid-tropospheric
7 CO2 are retrieved using the Vanish Partial Derivative Method (VPD) [Chahine et al.,
8 2005; Chahine et al., 2008; Olsen et al., 2009]. The sensitivity function of AIRS mid-
9 tropospheric CO2 peaks between 500 hPa to 300 hPa. AIRS mid-tropospheric CO2 is
10 retrieved globally in the middle troposphere day and night under clear and cloudy
11 conditions. Validation by comparison to in situ aircraft measurements and retrievals by
12 land-based upward looking Fourier Transform Interferometers demonstrates that AIRS
13 CO2 is accurate to 1-2 ppm between latitudes 30°S and 80°N [Chahine et al., 2008; Olsen
14 et al., 2009]. The mid-tropospheric CO2 retrieved via the VPD method captures the
15 correct seasonal cycle compared with those from Comprehensive Observation Network
16 for Trace gases by AIrLiner (CONTRAIL) [Chahine et al., 2005; Olsen et al., 2009].
18 3. Results and Discussions
20  In order to investigate the interannual variability of CO2, we first remove linear trend
21 from the data at each latitude band. Mean value has been added back to the detrended
22 data. The interannual variability of CO2 in the tropics and the polar region will be
23 discussed in sections 3.1 and 3.2, respectively.
25 3.1 Interannual Variability of Tropical AIRS CO2
26  We will explore the interannual variability of tropical CO2 and the influence of
27 ENSO on CO2 in this section. ENSO is the most important large-scale climate variability
28 in the tropics. The Southern Oscillation Index (SOI), shown in Fig. 1a, is an index for
29 ENSO. It is defined as the monthly mean sea level pressure difference between Tahiti and
30 Darwin. Negative (positive) SOI index corresponds to an El Niño (La Niña) event. We
31 have derived the spatial distributions of AIRS detrended CO2 in Feb 2005 (El Niño event;
1 Negative SOI Index) and in Feb 2008 (La Niña event; Positive SOI index). Figs 1b and
2 1c show the spatial patterns of the detrended AIRS CO2 for these two events.
4  During an El Niño event, warm sea surface temperature (SST) anomalies appear in
5 the central and eastern Pacific and cold anomalies appear in the western Pacific, and the
6 convection will move eastward to the central Pacific [Gage and Reid, 1987]. As a result,
7 convection in the central Pacific brings surface high concentration CO2 into middle
8 troposphere. In Fig. 1b, AIRS detrended CO2 data indicate that surface layer CO2 has
9 been lifted into the middle troposphere over the central Pacific region during the El Niño
10 event. Low concentration of mid-tropospheric CO2 is seen in the western and eastern
11 Pacific Ocean, where there is sinking motion of the Walker Circulation.
13  In contrast, during a La Niña event (Positive SOI index), the SST is warmer in the
14 western Pacific and its associated convection is stronger. In Fig. 1c, AIRS detrended CO2
15 data suggest that CO2 has been lifted into the middle troposphere over the western
16 Pacific. Lower mid-tropospheric CO2 is seen in the eastern Pacific Ocean as a result of
17 transport of low CO2 from high altitude to the middle troposphere. Fig. 1d presents the
18 difference of mid-tropospheric CO2 between El Niño and La Niña events. The difference,
19 (El Niño - La Niña), of AIRS mid-tropospheric CO2 is about 12 ppm over the central
20 Pacific and -1 to -2 ppm over the western Pacific.
22 3.2 Impact of Polar Vortex on AIRS CO2
23  The annular modes are the most important source of climate variability in the high
24 latitudes. To investigate the influence of the polar vortex on AIRS mid-tropospheric CO2,
25 we averaged the detrended AIRS polar CO2 north of 60°N from November to April, for
26 the years 2003 through 2007. The result is shown as solid line in Fig. 2a. The strength of
27 the polar vortex is characterized by the northern annual mode (NAM) index, defined as
28 the leading time series for the sea-level pressure anomalies within November to April
29 from 20N to 90N [Thompson and Wallace, 1998]. The detrended and inverted NAM
30 index is shown as the dashed line in Fig. 2a. Pearson’s correlation coefficient for the
31 detrended AIRS polar CO2 and detrended inverted NAM index is 0.7, and the
1 corresponding significance level is 9%. The significance statistic for correlations was
2 generated by a Monte Carlo method [Press et al., 1992; Jiang et al., 2004].
4  Fig. 2a indicates that there are strong polar vortices in 2005 and 2007, i.e., for those
5 years the NAM index is positive. When the polar vortex is strong, there is less horizontal
6 mixing of air between the mid-latitudes and high-latitudes. As a result, the transport of air
7 containing a high concentration of CO2 from the mid-latitudes into the polar region is
8 weakened. During those years the concentration of AIRS CO2 in the polar region should
9 therefore be relatively low. Fig. 2b shows that the mean of AIRS detrended polar CO2
10 from November to April in 2005 and 2007 was reduced.
12  On the other hand, Fig. 2a indicates that the polar vortices are weak in 2006 and
13 2008; the NAM index is negative in those years. For the case of a weak polar vortex, the
14 horizontal mixing is strong and mid-latitude air containing higher concentrations of CO2
15 can be transported into the polar region. Fig. 2c shows that the AIRS detrended polar CO2
16 was enhanced from November to April in 2006 and 2008. The difference of AIRS polar
17 CO2 between the strong polar vortex and weak polar vortex years is shown in Fig. 2d.
18 There is less CO2 at the polar region for the strong polar vortex years. The minimum
19 value for the difference can be 2-3 ppm. Student-t test is used to calculate the statistical
20 significance of the difference for CO2 concentration in the strong and weak vortex years.
21 The result is statistically significant when t is larger than a certain value t 0. The number
22 of degrees of freedom for the CO2 difference between two groups is equal to 2. Given the
23 number of degrees of freedom, t0 can be found from the t distribution table. t0 with a 10%
24 significance level is 2.9. When the t-value in Fig. 2e is larger than 2.9, the results is
25 within 10% significance level. The results obtained here are similar to those obtained for
26 stratospheric O3 in the northern polar vortex [Jiang et al. 2008a], where the principal
27 component PC1 of O3 is correlated with the NAM index at 100 hPa with a correlation
28 coefficient of -0.53. The underlying physics is the same, as a stronger vortex inhibits
29 transport of O3 from the lower latitudes into the polar region.
1 4. Conclusions
2  AIRS mid-tropospheric CO2 retrievals have been used to investigate the
3 interannual variability of CO2 in the middle troposphere for the first time. Global CO2
4 retrievals offer a unique opportunity to explore the relationships between the mid-
5 tropospheric CO2 concentration and large-scale atmospheric processes. Our analysis
6 suggests that the influences of El Niño events and polar vortex on the CO2 concentration
7 are apparent in the AIRS data. During El Niño, mid-tropospheric CO2 is enhanced in
8 central Pacific Ocean and diminished in the western Pacific Ocean. In the polar region,
9 mid-tropospheric CO2 is diminished if the polar vortex is strong and enhanced if it is
10 weak. It remains a challenge for GCMs to simulate these results.
14 This work was carried out at the Jet Propulsion Laboratory, California Institute of
15 Technology, under contract with the National Aeronautics and Space Administration.
2 Aumann, H. H., et al. (2003), AIRS/AMSU/HSB on the Aqua mission: design, science
3 objectives, data products, and processing systems, IEEE Trans. Geosci. Remote
4 Sens., 41, 253.
5 Bacastow, R. (1976), Modulation of atmospheric carbon dioxide by the Southern
6 Oscillation, Nature, 261, 116-118.
7 Bacastow, R., J. Adams, C. Keeling, D. Moss, T. Whorf, and C. Wong (1980),
8 Atmospheric carbon dioxide, the southern oscillation, and the weak 1975 El Niño,
9 Science, 210, 66-68.
10 Chahine, M., C. Barnet, E. T. Olsen, L. Chen, and E. Maddy (2005), On the
11 determination of atmospheric minor gases by the method of vanishing partial
12 derivatives with application to CO2, Geophys. Res. Lett., 32,
14 Chahine, M., et al. (2008), Satellite remote sounding of mid-tropospheric CO2, Geophys.
15 Res. Lett., 35, doi:10.1029/2008GL035022.
16 Dickinson, R. E., and R. J. Cicerone (1986), Future global warming from atmospheric
17 trace gases, Nature, 319, 109-115.
18 Feely, R. (1987), Distribution of chemical tracers in the eastern equatorial Pacific during
19 and after the 1982-1983 El Niño/Southern Oscillation event, J. Geophys. Res., 92,
21 Feely, R., et al. (1997), Variability of CO2 distributions and sea-air fluxes in the central
22 and eastern equatorial Pacific during the 1991-1994 El Niño, Deep Sea Res., Part
23 II, 44, 1851-1867.
24 Feely, R., et al. (2006), Decadal variability of the air-sea CO2 fluxes in the equatorial
25 Pacific Ocean, J. Geophys. Res., 111, doi:10.1029/2005JC003129.
26 Francey, R., et al. (1995), Changes in oceanic and terrestrial carbon uptake since 1982,
27 Nature, 373, 326-330.
28 Gage, K. S., and G. C. Reid (1987), Longitudinal variations in tropical tropopause
29 properties in relation to tropical convection and El Niño-Southern oscillation
30 events, J. Geophys. Res., 92, 14197-14203.
31 Inoue, H. Y., and Y. Sugimura (1992), Variations and distributions of CO2 in and over
32 the equatorial Pacific during the period from the 1986/88 El Niño event to the
33 1988/89 La Nina event, Tellus Series B, 44, 1-22.
34 Keeling, C. D., T. P. Whorf, M. Wahlen, and J. Vanderplicht (1995), Interannual
35 extremes in the rate of rise of atmospheric carbon dioxide since 1980, Nature,
36 375, 666-670.
37 Jiang, X., C. D. Camp, R. Shia, D. Noone, C. Walker, and Y. Yung (2004), Quasi-
38 biennial oscillation and quasi-biennial oscillation-annual beat in the tropical total
39 column ozone: A two-dimensional model simulation, J. Geophys. Res., 109,
41 Jiang, X., S. Pawson, C. D. Camp, E. Nielsen, R. Shia, T. Liao, V. Limpasuvan, and Y.
42 L. Yung (2008a), Interannual variability and trends in extratropical ozone. Part I:
43 Northern hemisphere, Journal of the Atmospheric Sciences, 65, 3013-3029.
44 Jiang, X., S. Pawson, C. D. Camp, E. Nielsen, R. Shia, T. Liao, V. Limpasuvan, and Y.
45 L. Yung (2008b), Interannual variability and trends in extratropical ozone. Part II:
46 Southern hemisphere, Journal of the Atmospheric Sciences, 65, 3030-3041.
1 Jiang, X., M. T. Chahine, E. T. Olsen, Q. Li, L. Chen, M. Liang, and Y. L. Yung (2009),
2 A study of polar stratosphere-troposphere exchange using AIRS CO2, Submitted
3 to Journal of Atmospheric Sciences, 2009.
4 Jones, C. D., M. Collins, P. M. Cox, and S. A. Spall (2001), The carbon cycle response to
5 ENSO: A coupled climate-carbon cycle model study, Journal of Climate, 14,
7 Meloen, J., P. Siegmund, P. van Velthoven, H. Kelder, M. Sprenger, H. Wernli, A.
8 Kentarchos, G. Roelofs, J. Feichter, C. Land, C. Forster, P. James, A. Stohl, W.
9 Collins, and P. Cristofanelli (2003), Stratosphere-troposphere exchange: A model
10 and method intercomparison, J. Geophys. Res., 108, doi:10.1029/2002JD002274.
11 Olsen, E. T., M. T. Chahine, L. L. Chen, X. Jiang, T. S. Pagano, and Y. Yung (2009),
12 Validation of six years of mid-tropospheric CO2 data from the Atmospheric
13 Infrared Sounder, Submitted to J. Geophys. Res..
14 Press, W., S. Teukolsky, W. Vetterling, and B. Flannery (1992), Numerical Recipes in
15 Fortran 77: The Art of Scientific Computing, 2nd ed., 933 pp., Cambridge Univ.
16 Press, New York.
17 Thompson, D. W. J., and J. M. Wallace (1998), The Arctic Oscillation signature in the
18 wintertime geopotential height and temperature fields, Geophys. Res. Lett., 25,
20 Wong, C. S., et al. (1993), Changes in equatorial CO2 flux and new production estimated
21 from CO2 and nutrient levels in Pacific surface waters during the 1986/87 El
22 Niño, 45, 64-79.
1 Figure 1: (a) Southern Oscillation Index (black line), El Niño and La Niña events are
2 marked by red dotted lines, (b) AIRS detrended CO2 in Feb 2005, (c) AIRS detrended
3 CO2 in Feb 2008, and (d) AIRS detrended CO2 difference between Feb 2005 and Feb
4 2008. Units are ppm.
1 Figure 2: (a) AIRS detrended CO2 (solid line) averaged from 60N to 90N and
2 detrended NAM index (dash line, inverted). (b) AIRS detrended CO2 average of 2005
3 and 2007, which are two strong polar vortex years. (c) AIRS detrended CO2 averaged in
4 2006 and 2008, which are two weak polar vortex years. (d) Difference of AIRS CO2 for
5 strong and weak polar vortex years. (e) t-value for the CO2 difference. All CO2 data are
6 averaged from November to April.