A Comparison of CO₂ and ¹³/¹²C Seasonal Amplitudes in the Northern Hemisphere

T. P. WHORF, C.D. KEELING, AND M. WAHLEN

Scripps Institution of Oceanography, La Jolla, California 92093-0220

INTRODUCTION

To better understand the sources and sinks of atmospheric carbon dioxide, the Scripps Institution of Oceanography (SIO) continues to maintain cooperative programs of CO₂ measurements with CMDL at Mauna Loa Observatory, Hawaii (MLO, 20°N), Point Barrow, Alaska (BRW, 71°N), Cape Kumukahi (KUM, 20°N), Samoa (14°S), and the South Pole (SPO), where air samples have been collected in 5-L glass flasks on a weekly to twice monthly basis, the latter at SPO. In addition, SIO continues to record atmospheric CO₂ concentrations on a continuous basis at Mauna Loa using a non-dispersive infrared (NDIR) gas analyzer installed on site in 1958. Studies of the MLO record in the last few years include a report on changes in the rate of rise of CO₂ and its relationship with global temperatures [Keeling et al., 1995].

More recently, seasonal cycle studies of data from MLO and BRW, augmented by data from Alert, N.W.T (ALT, 82°N), have been published [Keeling et al., 1996] that have yielded evidence of climate induced CO₂ changes in the form of possible changes in the growth of plants on a very large spatial scale. The Alert measurements have been collected under a cooperative program between SIO and the Atmospheric Environment Service of Canada. The SIO carbon dioxide program also continues to monitor CO₂ concentrations at Christmas Island (2°N), Baring Head, New Zealand (41°S), Raoul Island, Kermadec Islands (29°S), and La Jolla, California (LJO, 33°N), the last of which started in 1969 and now includes both continuous and flask measurements.

Since 1978 the ¹³C/¹²C isotopic ratio of atmospheric CO₂ has been determined from the same 5-L flask samples of CO₂ that have been measured for CO₂ concentration. In the early years isotopic ratio was measured at MLO, SPO, and LJO and by the mid-1980s at a total of ten sites. Measurements before 1992 were determined using a VG Micromass 903 and a VG SIRA 9 mass spectrometer at the Groningen Isotopic Physics Laboratory in the Netherlands [Keeling et al., 1989]. Samples since then have been analyzed using a VG PRISM Series II mass spectrometer at SIO.

Changes in the amplitude of the seasonal cycle in CO₂ concentration at MLO were first observed and reported in the early 1980s [Bacastow et al., 1985]. A more recent analysis of seasonal cycle variations observed at MLO has shown an increase in amplitude of 20% since the mid-1960s, an earlier drawdown of CO₂ in the late spring and early summer by up to a week since the late 1970s, and a correlation between amplitude changes there and tempera-tures averaged over the northern hemisphere [Keeling et al., 1996]. Here we discuss seasonal changes in the ¹³C/¹²C isotopic ratio of atmospheric CO₂ and how they compare with changes in amplitude of CO₂ concentration.

DATA AND ANALYSIS

Measurements of ¹³C/¹²C used in this study come from the same 5-L flask samples that have all been analyzed for CO₂ concentration with an NDIR analyzer of the same design as that installed at MLO. Concentrations were calibrated with our standard suite of reference gases and expressed in the X93 mole fraction scale. Air samples were rejected if pairs did not agree within 0.40 ppm of the lowest flask average (0.60 ppm at Point Barrow), if they were single analyses, or if found to be outliers having a residual greater than three sigma from our smooth curve fit. This fit, described in Keeling et al. [1989], uses four harmonics to describe the seasonal variations plus a cubic spline [Reinsch, 1967] to describe the interannual variations. A linearly-increasing gain factor was incorporated into the fitting routine when it became apparent that there were significant increases in the seasonal amplitude. This same fitting procedure has been applied to the isotopic measurements.

Isotopic data are obtained from samples of CO₂ extracted cryogenically which have passed the above 0.40 or 0.60 ppm cutoff criterion. Data for MLO and BRW are shown in Figure 1 as monthly averages and expressed as the reduced isotopic ratio, ¹³d, of atmospheric CO₂ [Keeling et al., 1989, p. 170]. Similar data have been obtained at the other SIO site locations. In 16 years of MLO ¹³ d isotopic data, over 700 daily samples have been analyzed amounting to approximately 44 per year, while at BRW in 14 years, some 500 daily samples have been analyzed yielding about 36 per year. Since 1992 when samples were first analyzed at SIO, gas standards have also been run daily in order to track any changes in the spectrometer output and thereby correct for them. These standards were recently calibrated against NBS19 (in 1996), to finalize calibrations begun in 1994. Along with this set of recent calibrations, a correction of +0.095 per mil was applied to all data analyzed at Groningen and previously published [Keeling et al., 1989; 1995]. This correction term is based on extensive duplicated measurements of these daily gas standards on the mass spectrometers at the Netherlands and at SIO.
 

Figure 1. Time trend of the reduced isotopic ratio, ¹³d, of atmospheric CO₂ in per mil difference from the international carbonate standard, PDB, for Point Barrow (BRW) and Mauna Loa (MLO). The data are shown as monthly averages and the curve by a four-harmonic fit with spline, as described.
 

RESULTS

Figures 2-4 show relative seasonal amplitudes computed annually by least-square fitting the annual function g(1 + At_m) multiplied by the four harmonics derived in the overall fit to the data of each year independently. In this expression, g denotes the annual relative amplitude, t_m the midpoint time of the record, and A, a constant gain factor for the record being fit expressed in percent change in amplitude per year. The harmonic function is thus phase locked to the fit of the overall data set. Amplitudes are expressed relative to the mean annual cycle arbitrarily set at the midpoint in time of the fit [Bacastow et al., 1985]. For purposes of intercomparison between stations or different time periods, the gain factor has been afterwards adjusted to a common date (see below).
 

Figure 2. Comparisons of relative seasonal amplitudes for CO₂ concentration and ¹³d at the three most northerly stations (a) Alert (ALT), (b) Point Barrow (BRW) and (c) La Jolla (LJO), each referred to the midpoint of its isotopic record (open circles). The large solid circles for LJO are continuous data.

Figure 3. Comparisons of relative seasonal amplitudes for CO₂ concentration and ¹³d at (a) Mauna Loa (MLO), flask data, and (b) Cape Kumukahi (KUM), each referred to the midpoint of its isotopic record (open circles). The large amplitude at KUM in 1984 is partly due to a strong shift in trend following the 1983 El Niño.^.
 
 

Figure 4. Comparison of seasonal amplitudes of CO₂ concentration for MLO (continuous data) and KUM, referred in this figure to the midpoint of the 1958-1995 Mauna Loa data. KUM amplitudes are adjusted to have the same mean as MLO amplitudes for the first 10 years of the KUM record (1979-1988).
 

Figures 2 and 3 show comparisons of seasonal amplitudes for CO₂ and ¹³d for our northern hemisphere sites, (ALT, BRW, LJO, MLO, and KUM) in which the CO₂ station data have been fit over the same time periods as there exists data for ¹³d, typically beginning around 1980 except for ALT data, which begins later. This is so that the relative amplitudes for the two quantities are referenced to the same times to give an optimal comparison. Though the daily averages used in the isotopic fits are not identically the same as those used in the CO₂ fits, they number approximately 90% of the number for CO₂, except at La Jolla where it is about 60%. In addition, the figures showing BRW, LJO, and MLO flasks have CO₂ data extending back in time to before the beginning of the isotopic data. In these cases, fits to these longer records have yielded earlier CO₂ relative amplitudes which have subsequently been adjusted so as to be referenced to the same time as in the shorter fit and then included in these figures. Increasing amplitudes are more apparent in these longer records. Gain factors calculated for the longer records shown at these three sites and referenced to 1970 near the start of these fits are BRW: 1.03 ± 0.16% yr^-1, LJO: 1.21 ± 0.17% yr^-1, and MLO flasks: 0.82 ± 0.15% yr^-1. The slightly larger fluctuations before and including 1981 for LJO and MLO flasks are attributed to there being significantly fewer annual data than in the latter part of each of these records. As an exception, the LJO record includes continuous CO₂ data in 1973-1975 (three large symbols) based on 5-hour averages of steady data subsequently converted to weekly data, so that these years are of significantly higher quality than other early years containing sparse flask data.

The fluctuations in ¹³d seasonal amplitudes occurring over several year periods show a strong tendency to follow the CO₂ fluctuations during the same times, especially at BRW and LJO. A significant part of semi-decadal to decadal changes in seasonal amplitude are likely due to large scale temperature changes [Keeling et al., 1996] while other fluctuations may be a result of the effect of interannual variations connected with El Niño events. Slightly larger isotopic swings in the annual amplitudes at LJO may be partly attributable to the fewer isotopic data at this site. Comparisons of the absolute seasonal amplitudes of ¹³d with CO₂ concentration for our northern hemisphere sites show a rate of change in ¹³ d with respect to CO₂ of close to 0.05 per mil per ppm, as would be expected by a predominantly terrestrial component in CO₂. An interesting but unexplained observation is that multi-year seasonal changes at ALT in both ¹³d and CO₂ seem to be out of phase with changes at BRW by a couple of years, particularly near 1992.

Comparison of seasonal changes in ¹³d and CO₂ concentration do not seem to follow each other as closely at MLO as at BRW and LJO. This may stem from part of the seasonal cycle in CO₂ concentration being produced by oceanic processes that influence it without significantly affecting the amplitude of ¹³d. A difference in the trends of the seasonal amplitudes in concentration and ¹³d is also apparent at MLO as evidenced by the differing gain factors ¹³d and CO₂.

Gain factors for CO₂ concentration and ¹³d, compared over the same time intervals and referred to 1970, are expressed as annual rates of change in Table 1. Annual rates referred to 1980 would be about 8-10% smaller than if referred to 1970. No significant change in the seasonal cycle of ¹³d is apparent in the MLO flask record since 1980, whereas the increase in seasonal amplitude of CO₂ concentration over the same period exceeds the standard error by more than a factor of two.

An additional observation of seasonal amplitude changes in CO₂ between MLO (continuous data) and KUM (Figure 4) where respective gain factors over the same period are shown in Table 1, shows no increase in seasonal amplitude at KUM since 1979. Also, no significant increase in amplitude is evident in the isotopic data. Since KUM, situated close to sea level, supplies principally oceanic air and MLO, at 3400 m, supplies air from aloft which is better mixed, some of this difference might stem from the sampling of different sources, MLO perhaps getting a greater percentage of air which has been influenced by passage over land.

TABLE. 1. Annual Rate of Increase in the Seasonal Amplitude of

Atmospheric CO₂ and ¹³d at Various Locations

*With reference to January 1, 1970, except for Alert referred to January 1, 1980.

Finally, there is a hint that whatever may be causing any divergence between the trend in seasonal amplitudes of ¹³d and CO₂ concentration at MLO may also be occurring at BRW and LJO as shown by lower gain factors for ¹³d over the same intervals (Table 1). These changes, however, are not significant since the isotopic records are not yet long enough to draw any firm conclusions.
 

CONCLUDING REMARKS

The SIO ¹³C/¹²C isotopic data have recently received final calibrations and have been analyzed to look for similarities and significant divergences in seasonal amplitude from the CO₂ concentration data. At several northerly sites including BRW and LJO, isotopic changes in seasonal amplitude appear to track the CO₂ amplitude changes quite well over periods of several years, suggesting that a dominant component in the seasonal signal of CO₂ is terrestrial plant activity. At other sites such as MLO in Hawaii, differences between trends in ¹³d and CO₂ concentration suggest an oceanic component. At this time, when subtle changes are just becoming apparent in concurrent concentration and ¹³C/¹²C records of atmospheric CO₂, we hope to be able to continue these measurements, as in the case of concentration data in the early 1980s when an increase in seasonal amplitude first became apparent in our longest records.

Acknowledgments. This work was supported by Grant ATM91-21986 of the National Science Foundation, Grant DE-FG03-95ER62075 of the U.S. Department of Energy and Grant NAGW-2987 of the U.S. National Aeronautics and Space Administration. In addition, through Contract 50RANR100090, NOAA provided for the cost of maintaining our primary standard reference gases used to maintain accuracy in our measurements of atmospheric CO₂. We are grateful to CMDL for providing assistance in our sampling program and for maintenance of our continuous recording instrument at MLO.
 

REFERENCES

Bacastow, R.B., Keeling, C.D., Whorf, T.P., Seasonal Amplitude Increase in Atmospheric CO₂ Concentration at Mauna Loa, Hawaii, 1959-1982, J. Geophys. Res., 90, 10,529-10,540, 1985.

Keeling, C.D., Bacastow, R.B., Carter, A.F., Piper, S.C., Whorf, T.P., Heimann, M., Mook, W.G., and Roeloffzen, H., A three-dimensional model of atmospheric CO₂ transport based on observed winds: 1. Analysis of observational data, in Aspects of Climate Variability in the Pacific and the Western Americas, edited by D. H. Peterson, Geophysical Monograph, American Geophysical Union, 55,165-236, Washington, DC, 1989.

Keeling, C.D., Whorf, T.P., Wahlen, M., van der Plicht, J., Interannual extremes in the rate of rise of atmospheric carbon dioxide since 1980, Nature, 375, 666-670, 1995.

Keeling, C.D., Chin, J.F.S., Whorf, T.P., Increased activity of northern vegetation inferred from atmospheric CO₂ measurements, Nature, 382, 146-149, 1996.

Reinsch, C. H., Smoothing by spline functions, Numerische Mathematik, 10, 177-183, 1967.
 

[back] [contents] [next]