Airborne observations of the tropospheric CO2 distribution and its controlling factors over the South Pacific Basin

. Highly precise measurements of CO 2 mixing ratios were recorded aboard both the NASA DC-8 and P3-B aircraft during the Pacific Exploratory Mission-Tropics conducted in August-October 1996. Data were obtained at altitudes ranging from 0.1 to 12 km over a large portion of the South Pacific Basin representing the most geographically extensive CO 2 data set recorded in this region. These data along with CO 2 surface measurements from the National Oceanic and Atmospheric Administration/Climate Monitoring and Diagnostics Laboratory (NOAA/CMDL) and the National Institute of Water and Atmospheric Research (NIWA) were examined to establish vertical and meridional gradients. The CO 2 spatial distribution in the southern hemisphere appeared to be largely determined by interhemispheric transport as air masses with depleted CO 2 levels characteristic of northern hemispheric air were frequently observed south of the Intertropical Convergence Zone. However, regional processes also played a role in modulating background concentrations. Comparisons of CO 2 with other trace gases indicated that CO 2 values were influenced by continental sources. Large scale plumes from biomass burning activities produced enhanced CO 2 mixing ratios within the lower to midtroposphere over portions of the remote Pacific. An apparent CO 2 source was observed in the NOAA/CMDL surface data between 15øN and 15øS and in the lower altitude flight data between 8øN and 8.5øS with a zone of intensity from 6.5øN to løS. Inferred from these data is the presence of a Southern Ocean sink from south of 15øS having two distinct zones seasonally out of phase with one another.


Introduction
"Fixed air," or carbon dioxide (CO2), was first noted in 1754 by physician Joseph Black [Jaffe, 1942], and the earliest records of atmospheric CO 2 measurements are described in the nineteenth century literature [Callendar, 1958]. However, prior to the late 1950s, the atmospheric concentration of CO 2 was thought to exhibit significant variability until precise and accurate measurements utilizing infrared absorption and manometric techniques were instituted by C. D. Keeling. As an offshoot of this initial work, an extensive global surface network currently exists for measurements of CO 2 and its isotopes yet, despite these efforts, much remains uncertain in terms of the global carbon budget.
Contributing to these uncertainties are unprecedented human disturbances of the atmosphere predominantly in the form of fossil fuel combustion and, to a lesser degree, land use changes. In 1995, for example, the global annual CO 2 emission from the burning of fossil fuels, including gas flaring, and hydraulic cement production was estimated at 6.4 GtC ( Their results suggest that this trend is partially attributable to human activities. Clouding the picture is the increasing difficulty in separating out changes that are part of short or long term natural trends from those which are anthropogenically induced. What is 5664 VAY ET AL.' AIRBORNE OBSERVATIONS OF THE TROPOSPHERIC CO 2 DISTRIBUTION CO 2 were undertaken in the South Pacific Basin; a region relatively uncharted above the planetary boundary layer. Prior aircraft measurements in this portion of the southern hemisphere were focused over the Australian-New Zealand region and south to Antarctica [Pearman and Beardsmore, 1984], Japan to Australia [Nakazawa et al., 1991;Matsueda and Inoue, 1996], and the United States to Tahiti . The following text therefore describes the first extensive airborne CO 2 data set encompassing the breadth of the basin. These data provide valuable information on the spatial distributions of CO 2 which can be used to develop, test, and enhance the predictive abilities of 3-D global CO 2 models and foster a better understanding of the globai carbon cycle.

Experiment
The Pacific Exploratory Mission-Tropics (PEM-T) was designed as an intensive airborne campaign over the South Pacific Basin with the primary objectives of providing baseline data on the chemical species important in controlling the oxidizing power of the troposphere and the factors governing their concentrations. Since the atmosphere in this remote geographic region is relatively unexplored and possibly the most pristine globally, a background level assessment of these chemical species was considered crucial before anthropogenic inputs increase further from expanding economies and population growth.
PEM-T coincided with the austral late winter/early spring, when biomass burning reaches its seasonal maximum (August-October) in the southern tropics [Kirchhoff et al., 1996] nominal airspeed of 225 ms -1 at cruise altitude, a flight duration of 12 hours, and an altitude range of 0.3 km to 12.8 km. Actual flight tracks for the DC-8 during the mission are presented in Figure 1 whereas those for the P3-B are shown in Figure 2. Flight patterns typically consisted of constant altitude legs, ramps, spirals, and additionally in the case of the P3-B, Lagrangian circles. Bases of operation were located at Hawaii (Oahu), Christmas Island, Tahiti, Fiji, New Zealand (Christchurch), Easter Island, and Ecuador (Guayaquil) and provided excellent opportunities for good aerial spatial coverage of the basin longitudinally from 170øE to 80øW as well as meridionally from 20øN to 72øS. A listing of the science flights used in this subsequent analysis can be found in Table 1

C02 (ppmv)
Plate 3. Regional distribution of CO 2 during PEM-T for (a) 0-2 km, (b) >2-6 km, and (c) >6-12 km altitude ranges. Data were grouped into 5 ø x 5 ø bins and then averaged.  , 1989]. At approximately 2 hour intervals, the instrument sensitivity or "span" was verified by flowing a second standard (span gas) through the LI-COR sample cell having a CO 2 mixing ratio 10-15 ppmv higher than the reference gas. This span gas was a synthetic air mixture purchased from Scott Marrin  Ranges comprise geographic areas where CO 2 data are available.    . A similar feature is also noted in the PBL and lower troposphere data however, the peak begins at 10øN. Pertinent to these observed increases is the location of the ITCZ at 8ø30'N.   In the SH, the mid to upper troposphere appears relatively rich in CO 2 compared to the lower troposphere owing to the CO 2 sink at the ocean surface, but also because the vertical distributions of SH CO 2 are controlled by top-down diffusion resulting from the between 30øS and 60øS. They also note recent shipboard meas-interhemispheric exchange process that occurs more frequently . Numerical experiments on tracer particle transport have demonstrated that this process is most pronounced in the NH summer [Nikaidou, 1989]. This effect is illustrated at 40øS to 60øS in Plate 5 where CO 2 concentrations in the 6-10 km band exceed those found in the lower troposphere. These levels most likely reflect CO 2 mixing ratios from the previous NH nongrowing season (winter 1995), whereas the lower values in the 10-12km band can be attributed to the early NH spring (1996). This is further substantiated in viewing Plate 1 where the next wave of NH air is observed propagating southward, being carried further aloft at the ITCZ by deep convection, and then progressing into the midsouthern latitudes at the higher altitudes. At 46øS, this

NH early spring air is observed both at higher levels and diffusing top-down into lower layers mixing with air from the previous NH winter. Aircraft measurements conducted in the Australian-New
Zealand region by Pearman and Beardsmore [1984] showed a midtroposphere (3.5 to 5.5 kin) concentration maximum in October which lagged behind the upper troposphere (>8.5 kin) by a month and attributed this to high altitude meridional transport of NH air into the SH. More recently, however, Tans et al. [1989] have proposed that the seasonal cycle at 40øS cannot be entirely ascribed to transport from the NH rather a "tongue" of seasonality northern portion of that continent. Measurements made south of 25øS, were obtained in an air mass originating beyond 60øE that clipped the southern part of Australia as it entered a trough over New Zealand. Observed elevated levels of C2H 4, having a x-2 days for this latitude and season, along with wind speeds at flight level suggest that the plumes sampled south of 25øS originated in Australia rather than Africa.
The highest CO 2 mixing ratios were observed within biomass burning plumes residing below 5 km off the western South American coast during P3-B flight number 17. Levels of exists near the surface from 40øS to 10øS, underneath the mid tro-ACO 2 within the plumes were -4 ppmv, whereas ACO, AC2H 6, posphere, with its seasonality 100ø-180 ø out of phase with higher ACH3C1 were markedly higher being -300ppbv, -2000pptv, and altitudes. Plate 1 indicates that this may indeed be the case except -175 pptv, respectively. A corresponding C2H2/CO ratio of the tongue appears to extend beyond 40øS. Tans etal.
[1989] note -2pptv/ppbv for all cases indicates these plumes were less that this surface layer progressed northward, while the midtro-processed and contained fairly recent emissions. Ten-day back pospheric seasonal signal originates from the NH. In examining Plate 5 at latitudes from 15øS to 40øS, the lower tropospheric CO 2 mixing ratios are observed to exceed those in the mid to upper troposphere. This was due in part to the influence of plumes, most likely from biomass burning, residing within the 2-6 km band encountered during the transit flight from New Zealand to Fiji (J. Logan, personal communication, 1997). We note that at the higher southern latitudes all altitude bands exhibit similar CO 2 concentrations indicating a zone of well mixed air probably owing to recent winter convection.

Biomass Burning
Although the CO 2 spatial distribution in the South Pacific Basin was significantly influenced by interhemispheric transport, regional processes also played a Since biomass burning tends to follow a seasonal pattern, a totally different picture of the basin may be observed in subsequent seasons. During the PEM-T timeframe, however• we find that it plays a significant role in affecting the spatial CO 2 distributions.

3.4ø Interhemispheric Transport
Above the Earth's surface, CO 2 in the atmosphere is unreactive and therefore can function as a conservative tracer for elucidating atmospheric mixing processes [Bolin andKeeling, 1963]. Plates 1 and 2 are suggestive of NH air depleted in CO 2 propagating southward; however, other tracer species can be used in conjunction with the CO 2 to substantiate the case of NH/SH exchange. DC-8 flight number 8, flown due north out of Easter Island, provides an excellent example of interhemispheric transport occurring in the eastern portion of the South Pacific Basin and the use of CO 2 along with other trace species for verification. A descending vertical profile was performed at the northernmost point of the flight at 7ø15'S and is shown in Plate 7 for the trace species CO2, CO , and CH 4. Above 7 km, CO 2 shows a decreasing trend while CO and CH 4 exhibit elevated levels; evidence of NH air. Accompanying these increases are enhanced levels of C2H 6, C3H 8, C2H 2, 'CH3C1, PAN, fine and ultra fine aerosols, HCFC 14lb, HCFC 142b, and HFC 134a. In particular, the HCFC 14lb displayed a ! pptv enhancement corresponding to an air mass "age" of about 6 to 8 months (E. Atlas, personal communication, 1997). The C2H2/CO ratio was approximately 0.5 pptv/ppbv indicative of an aged air parcel having no inputs of fresh emissions from combustion sources for at least a week [Smyth etal., 1996]. Noteworthy is a layer with similar chemical signatures occurring between 1.5 and 4 km. As shown in Plate 8, within the layers of low CO 2 values the wind is due north or northeast, while elevated levels are measured when winds are from the south or southeast. Meteorological analysis shows southeast flow skirting along and off the coastal area of westcentral South America, whereas the southern flow is from the western Pacific Ocean. During periods of low CO 2 mixing ratios, a northern hemispheric flow was clearly indicated that emanated from Central America.

Upwelling
A survey of data collected from 1958 to 1962 revealed a natural release of CO 2 at the ocean surface in the tropical oceanic areas [Bolin and Keeling, 1963]. Subsequent surface measurements of 13C/12C ratios have demonstrated that the source can be attributed to the release of CO 2 from seawater [Keeling and Carter, 1984] along the equator where the north and south equatorial currents from each hemisphere meet causing a divergence or upwelling. Indications are the observed atmospheric peak is a stationary source (except possibly during E1 Nifio events) resulting from waters supersaturated in CO 2. Several authors have since noted this equatorial source at varying degrees of latitude and source strengths, for example, 12øN to 12øS, 1.3 GtC [Heimann and Keeling, 1989] and 8.5øS with a zone of intensity from 6.5øN to løS. Interestingly, these elevated CO 2 concentrations at the lower altitudes appear confined to a vertical column that is bounded on either side by the ITCZ and SPCZ suggesting that the enriched CO 2 column may actually shift in time with the seasonal movement of these zones. However, since CO 2 sources/sinks can only be inferred from these data, low CO 2 air traveling above the surface in the absence of strong surface fluxes must also be considered as a plausible explanation for this equatorial "source." Plate 1 also demonstrates the importance of vertical trace gas profiles, particularly in potential source/sink regions, as they provide an additional constraint for global scale trace gas budgeting models [Fung et al., 1991 ].
Upwelling from the depths of cold, mineral-rich, supersaturated CO 2 water also occurs on the eastern sides of oceans. South America has one of the most vigorous areas of upwelling in the world off its western coastline due to the southeasterly trade winds enabling the northward flowing Humboldt, or Peru, current to reach the surface. Enhancements in CO 2 mixing ratios of the order of 1.4 ppmv were observed during P3-B low-altitude flight legs (-0.1 krn) off the South American coast as shown in Plate 3a.
Accompanying these elevated CO 2 concentrations were notable levels of DMS which is produced by the metabolic processes in certain types of algae residing in the nutrient-rich waters therefore suggestive of an upwelling source. Other tracers such as CO and CH 4 demonstrated no departure from background levels during these low-altitude runs having concentrations of-60 and -1700 ppbv, respectively. Evidence of the upwelling signature in the free troposphere was dwarfed by the outflow from the continent's interior of biomass burning plumes containing high CO 2 mixing ratios.

Summary and Conclusions
High-precision CO 2 measurements were made over a significant portion of the South Pacific Basin from August 18 to October 5, 1996, as part of the PEM-T mission. These tropospheric measurements represent the most extensive aerial CO 2 data set in this region therefore greatly augmenting the existing database. Results from this rich data set offer insight into the source/sink processes occurring within this remote region. We find that interhemispheric transport plays a key role in determining the spatial distribution of CO 2 in the SH particularly at the upper levels. However, near the surface, from south of 10 ø S, there exists a "tongue" of air seasonally out of phase with the midtropospheric levels that cannot be attributed to transport from the NH. Biomass burning activities, producing elevated levels of CO 2, contributed significantly to the zonal inhomogenity observed around the basin perimeter. A CO 2 source was observed in the equatorial region, whereas a Southern Ocean sink beginning at 15øS was inferred from these data. These results will likely have significant implications for the development and testing of atmospheric global carbon cycle models. PEM-T occurred in the months preceding an E1 Nifio which is currently described as "the climatic event of the century." A follow-on mission is tentatively scheduled for the austral fall of 1999 during the anticipated waning months of this eve•_t.