A survey of continental concentrations of atmospheric CO in the southern hemisphere

~tntospheric

Emairtrnmenl Vol. 23, No. 2. pp. 461-466.

cmm4-69~1/89 s3.oo+o.oo Pergamon Ress plc

1989.

Printed in Great Britain.

A SURVEY OF CONTINENTAL CONCENTRATIONS OF AT~~SPHERIC CO IN THE SOUTHERN HE~ISPHE~E V. W.J.H.

KIRCHHOFF and E. V. A. MARINHO

Institute de Pesquisas Espadais-INPE, (First received 12 January

C.P. 51X%0 Jo& dos Campos, SP, Brazii

1988 and in final

form

3 August 1988)

Abstract-The first large scale survey of surface CO con~ntrations at Southern Hemisphere continental sites is described. Marine sites are compared to sites with a true continental character with the objective to identify different ecological surface conditions in terms of CO concentrations. The marine sites at the Atlantic coast show the lowest concentrations, about 100 ppbv, whereas the sites in the Savannah region show concentrations 3 times as large owing to the inffuence of nearby biomass burning activity. The observations were highly variable, with one result as high as 700 ppbv. These high values are comparable to sites near urban developments. Sites in the Amazonian rain forest show concentrations as low as the coastal sites, on the average, but sporadic peaks have been seen when air masses are brought in from city areas or from large forest fires. Key word index: Carbon monoxide, atmospheric CO.

ency induced by the trend in CH, (Khalil and Rasmussen, 1985). It has been estimated that CO levels increase yearly at a rate of about 2% (Rinsland and Levine, 1985) but long term measurements at fixed sites have not yet noticed such a trend. CO studies in the tropics are especially interesting since most biomass burning activities are thought to occur in tropical regions. The dry season in the low latitudes may be very long with virtually complete absence of any rain. The vegetation dries out and catches fire very easily and can propagate itself over long distances. This is the usual behavior, for example, of the Brazilian Cerrado region.

INTRODUCFION Carbon monoxide, CO, is one of the most important atmospheric minor constituents, being actively involved in many chemical reactions. In the troposphere it may represent the role of an ozone precursor. Starting with the reaction with OH, and after several intermediate catalytic reactions, (Crutzen, 1979), ozone may be produced photochemically in the troposphere. This scheme is especially important in the tropics, where the OH radical is produced at larger rates than at higher latitudes, because of the availability of higher intensities of ultraviolet (u.v.) radiation. CO may be produced from natural processes, but also from anthropogenic activity. The oxidation of HCs naturally emitted by vegetation, CJHs, and CIOH,,, and especially the oxidation of CH, represents the natural production mechanisms. Among man-made sources, there are industrial activities and biomass burning activities. Presently, it is generally believed that the anthropogenic sources are of the same order of magnitude as the natural sources, but it is not easy to obtain reliable estimates of the source strengths. This is especially true in South America where representative measurements are virtually nonexistent. It is also very likely that the anthropogenic sources are increasing in strength every year due to ever expanding industrial activites (about 5% a-’ in Brazil) and more demand for land use and the corresponding biomass burning (this rate of increase must be considered highly uncertain, but conservatively it should be at least as large as the above rate of 5% a- ‘). Evidently this would lead to a CO trend, with a tendency to decrease the tropospheric OH concentrations. Such behavior would increase the same tend-

PREVIOUSMEASUREMENTS Some measurements of CO have been made in the Southern Hemisphere at fixed marine stations from samples collected on shipboard or from airplanes (Seiler et al., 1984). Very little is known from truly continental sites especially in the tropical region. Brazil is the largest country in South America, extending east to west over almost 4000 km from 35”W to 7&W, and north to south over an almost identical length from 5”N to 34”s. It represents therefore, a natural choice for performing Southern Hemisphere tropical measurements. With an area of 8.5 x lo6 km’ it is still an ideal place to study the natural atmosphere at low latitudes and the influence of the many very distinct ecosystems of its territory. The general position of Brazil relative to South America is shown in Fig. la. Figure lb shows some of its ecosystems and the locations of some of our sampling sites. Previous measurements of atmospheric CO have been made in Brazil from surface samples collected at

461

462

V. W. J. H. KIRCHHOFFand E. V. A. MARINHO

m

Equatorial forest

m

Tropical forest

Fig. 1. (a) Map of South America, showing the position of Brazil, in scale; (b) surface ecosystems Brazil and position

sites and sent to the U.S. for analysis (Wofsy, 1989), and also during an expedition on board of an aircraft that flew over the Amazon and the Cerrado regions (Crutzen et al., 1985; Delany et al., 1985; Greenberg et al., 1984). temporary

MEASUREMENT

TECHNIQUE

The basic CO measurement technique used is chromatographic separation followed by mercury vapor detection (Seiler, 1974). This technique has much better chromatographic resolution and measurement precision, as well as linearity of response, as compared, for example with the FID technique, for low concentrations in the ppbv (parts per billion by volume) range. The carrier gas is dried and filtered to eliminate interference from reducing gases. SO, and HCs, which could react with mercuric oxide at the reaction chamber temperatures are eliminated on a routine basis since the instrument is dedicated to the measurement of CO only. The instrument is manufactured in the U.S. Samples are individually injected and the concentration is determined by comparison with calibration gases. Injection into the flow system is accomplished by initial transference to a 2 ml sample loop which is, after extensive flushing with sample air, introduced into the main flow using a six-port valve. The operation is very simple and fast and has excellent resolution and precision. Around 100 ppbv, typical precision is 0.2%. Each sample is analyzed in about 5 min. The samples are collected by a small portable air pump that pressurizes stainless steel cans of 800 ml volume, electropolished internally and especially manufactured for air sampling purposes, developed by R. Rasmussen, Oregon Graduate Center. For the present analysis all samples have been analyzed in the lab in less than a week’s time after sampling, and typically after 2 or 3 days. OLLJECTIVE OF PAPER

Our own measurement program and studies in atmospheric chemistry have been started in 1978 with

of sampling

in

sites.

an 0, measurement program conducted at Natal (6”S), where balloons and rockets have been launched to measure 0, (Kirchhoff et al., 1981; 1983; Kirchhoff, 1984; Logan and Kirchhoff, 1986). The CO measurement program is more recent, having been started in 1985. This is our first report on the CO resuits. We describe the first analyses of surface CO concentrations performed in our laboratory, in SIo Jose dos Campos, from samples collected at several stations in South America (Brazil). The sites were chosen for special ecological interests. Some of these are truly continental stations. We plan to continue CO analyses in the future, and expect that in due time it will be possible to learn about local sources and possible trends. This report describes a survey of the distribution of CO concentrations over some representative continental

areas in South America.

RESULTS

Samples have been collected at several different sites in Brazil in an attempt to characterize different ecosystems (Fig. lb). These sites belong to the following four different groups: the western group in the Amazon equatorial rain forest region with two sites, B&m (1.5”S,48.5”W)and Manaus (3.1”S, 6O.o”W); the Brazilian Cerrado (a type of Savannah, grassland with scattered small trees), a group with sites at Cuiabl (15.6”S, 56.1”W), the geodesic center of South America and Porto Jofre (17.3”S, 56.8”W), which was chosen exactly in the center of the so called Pantanal region, a low land area that is inundated every year in the rain season, a sui generis ecosystem that extends over 230 x lo3 km’; the third group is representative of the

463

Survey of concentrations of CO in Southern Hemisphere

r

south in the region of the country where industrial activity is largest, with sites at S&o Jest dos Campos (23.2”S, 45.8”W), Cachoeira Paulista (22.7”!$45.o”W) and Campos do Jordlo (22.7”S, 45.6”W). Sfo JO& dos Campos is the one closest to urban characteristics, Cachoeira is relatively more rural, whereas Campos do JordHo is a rural leisure and vacation resort at an altitude of 1700 m. The fourth group has two typically marine stations at Massaguacti (23.6”S, 45.2”W) and Natal (5.8”S, 35.2”W), both at the eastern Atlantic coast of Brazil. These marine stations have strong sea breeze wind regimes. It should be noted that we have always tried to sample clean air, that is, we have tried to avoid sampling close to pollution sources. The data collection time periods and respective averages are summarized in Table 1. Histograms for the CO concentrations have been prepared for the five sampling sites shown in Fig. 1, but results are also available for three additional sites, which will be discussed as well. The marine stations Natal and MassaguaCu show lowest concentrations. The largest was found for the urban site SHo Jest dos Campos, closely followed by the cerrado sites Cuiabi and Port0 Jofre in the Pantanal region. From the group of continental sites Belkm and Manaus show lowest concentrations.

NATAL ii = 105.8 p = 31.8 n= 27

CO Concentration [ppbv]

Fig. 2. Histogram of CO concentrations sampled

Natal

in Natal.

A marine site (Fig. 1) subject to strong prevailing winds from the sea. From a total of 27 samples the average concentration obtained is 105.8 ppbv. Not much scatter in the data can be seen (Fig. 2), with most samples falling in the interval 75-100 ppbv. The sampling period was September and October.

interval. The sample collecting February to August.

period

was from

Manaus

This data set from the Amazonian rain forest is the result of a two month measurement campaign that took place directly in the forest. Daily measurement routines produced the 36 daily averages shown in Fig. 4. Most points are within the 75-125 ppbv concentration interval, but high sporadic CO peaks have been detected. These peaks have been associated with unusual wind regimes which have transported air masses from city areas (Manaus) to the measurement site in the forest. Without this perturbation, however,

Belth

Samples from Belim (Fig. 3) were collected outside the downtown area, about 7 km southeastwards. Since the prevailing wind blows from the east the samples should represent clean air. Although Belim is close to the coast, with the easterly winds it may be expected to show some influence of the forest on the CO data. The average of 17 samples gave 168 ppbv of CO, with most samples within the lOG175 ppbv

Table 1. Sampling sites and time periods Location Bekm* Manaus* Natal* CuiabB* P. Jofre Cachoeira C. do Jordao S. J. Campos Massaguacu

Latitude South

Longitude West

Samples

Period

Average

CT

1.5 3.1 5.8 15.6 17.3 22.7 22.7 23.2 23.6

48.5 60.0 35.2 56.1 56.8 45.0 45.6 45.8 45.2

17 36 27 13 15 17 14 49 16

Feb 87-Aug 87 Apr 87-May 87 Sep 87-Ott 87 Jul87-Sep 87 Jul87 Mar 87-&p 87 Aug 87-Ott 87 Aug 87-Sep 87 Sep 87-Ott 87

168.4 108.4 105.8 317.6 262.4 195.2 195.8 331.3 107.0

99.7 19.4 31.8 197.5 60.1 84.7 61.6 176.2 43.1

* Sites shown on map in Fig. 1; concentrations shown in Figs 26.

V. W. J. H. KIRCHHOFF and E. V. A. MARINHO

464

i = 168.4 p = 99.7

n=

17

CO Concentration Fig. 3. Histogram

[ppbv]

of CO concentrations

sampled

in Bedim.

for the other sites. The average CO concentration, a large value of 317 ppbv, and the scatter of values, are consistent with biomass burning activities that are common in that area. We have investigated this area further, by also taking samples in a more remote area of difficult access, called the Pantanal. Distant about 245 km from Cuiaba, in the south, we have collected 15 samples in July. The average CO mixing ratio was 262 ppbv at the Porto Jofre site. This high CO prevailed in a regime of dry weather with clear skies when no agricultural fire could be noticed in the area.

MANAUS

Sdo JosP dos Campos

loo

I50

CO Concentration Fig. tr,

2

[ppbv:

Histogram of CO concenons sampled in Manaus.

the rain forest average CO concentration to the sites at the Atlantic coast.

is very close

Cuiabd

This site is representative of the Campo Cerrado ecosystem. The samples are taken far enough from the city limits so that city effects may be ruled out. Between July and September 13 samples were collected, shown in Fig. 5. The individual concentrations of the samples are much more scattered in this case than

The air samples were collected about 6 km, outside the downtown area. Four outsiders, at the high side, are not shown in Fig. 6. These 49 samples were obtained during August-September 1987, giving an average concentration of 33 1 ppbv. This average is the highest found and probably reflects the urban, industrialized character of this area. Samples taken farther away from this area, in a town about 100 km to the east, Cachoeira Paulista, with a much stronger rural character, seem to confirm this interpretation. The Cachoeira Paulista average for 17 samples in the March-September period was 195 ppbv. Still another site of this area, to the southeast is Massaguacu, on the Atlantic coast. Here the average concentration observed was 107 ppbv. The smallest CO concentration observed so far was collected at this site. To the north, in the mountains, at 1700 m of altitude is Campos do Jordao, a resort area with almost no industrial activity. The average of 14 samples was 196 ppbv, almost twice as large as the coastal sites.

DISCUSSION

The sites at the Atlantic coast, Natal and Massaguacu, show the lowest CO concentrations of our

Survey of concentrations

of CO in Southern Hemisphere

465

3

CUlABi\

0

IX)

zoo

300

450

500

700

7

CO Concentmtion [ppbv]

Fig. 5. Histogram of CO concentrations sampled in CuiabL.

CO Concentmtlon[ppbv]

Fig. 6. Histogram of CO concentrations sampled in SHo Jo& dos Campos.

The results of our survey clearly show a tendency for larger CO concentrations over the continent as compared to marine sites. In large part, this seems to be the result of biomass burnings previously observed in the NCAR aircraft expedition of 1979,198O (Greenberg et al., 1984; Delany et al., 1985; Crutzen et al., 1985). This first insight into the CO concentration dist~bution between marine and continental sites in South America was a major outcome of that field campaign, which took place in the dry season. As part of our results, we have shown surface CO concentrations in the Amazonian rain forest during the wet season, with concentrations 3 to 4 times lower than the dry season con~ntrations. This result implies a large seasonal variation of the CO concentration, with strong effects on the whole photochemistry of the lower atmosphere.

CONCLUSIONS

survey, about 107 ppbv, comparable only to values also seen in the rain forest. The urban area group of sites gave average CO concentrations between 2 and 3 times as large, which is also true for the Cerrado group CuiabP and Porto Jofre. The results for BelCm, part of the western Amazonian rain forest group, was much higher than the Manaus result, presumably owing to contamination from the big city complex, despite our effort to sample clean air. Our data base is still too small to consider seasonal effects in detail. It is possible, however, that our values a’re on the high side. The averages from the Atlantic coast, collected during September and October, are more than twice as large as averages observed in Australia, 44 ppbv (Fraser et al., 1986), and this value is based on the same calibration scale. For a station in South Africa, Seiler et al. (1984) report values between 53 and 80 ppbv, which at the Rasmussen scale should be reduced by about 30%, and ovlr the Pacific, Heidt et al. (1980) report about 50ppbv of CO. It appears, therefore, that the CO concentrations measured in Brazil are considerably higher than the values reported for Australia and South Africa, even taking into account a seasonality factor. Large 0, concentrations have also been reported for Natal (Logan and KirchholI, 1986).

We describe results of the first large scale survey of surface CO concentrations at remote continental sites in Brazil. Marine sites and regions close to urban developments were sampled for comparison. The Atlantic coast sites show the lowest averages of about 107 ppbv CO. Compared to resuhs obtained in the Pacific region, these values are at least twice as large. Our results for the geodesic center of South America, in Cuiaba and the Pantanal region show averages of 317 and 262 ppbv for CO, comparable to the urban sites SBo Jose dos Campos (331 ppbv), Cachoeira Paulista (195 ppbv) and Campos do JordZo (195 ppbv). A relative comparison of our major sampling sites is shown in Fig. 7. We report the first set of measurements made inside the Pantanal region, at Porto Jofre. It is important to note that neither fires nor smoke were noticed in the area during the sampling period. Long range transport of CO enriched air may be responsible for bringing air from large development areas in the north-northwest of the state. An interesting implication of our results is a probable large seasonal (dry, wet) variation of the CO con~ntration in the Am~onian rain forest area. Our averages of about 100 ppbv obtained in the wet season

V. W. J. H. KIRCHHOFFand E. V. A. MARINHO

466

0 NATAL

MANAUS

BELEM

CUlPgA

S JOSE

Sampling sites Fig. 7. Relative average concentrations of CO comparison of Brazilian sites.

are 3 to 4 times smaller than typical reported by other groups.

for

dry season values

Acknowledgements-We are grateful to Rei Rasmussen for lending us the air sample collecting system and for the supply of calibration gas. We also acknowledge calibration gas obtained from Wolfgang Seiler. We thank Steven Wofsy for suggesting the purchase of the CO instrument and for supplying us with replacement parts. Maria Angklica has performed most of the laboratory analyses and Francisco Mesquita has helped us with laboratory maintenance. Special thanks are due to the sample collecting team at the stations: Nilson Luiz in Cachoeira Paulista, Isa da Silva in Belbm, Adilson Palone in CuiabB, and Jo& Alves in Natal.

REFERENCES

Crutzen P. J. (1979) The role of NO and NO, in the chemistry of the troposphere and stratosphere. Ann. Rev. Earth Planet. Sci. I, 443-472. Crutzen P. J., Delany A. C., Greenberg J., Haagenson P., Heidt L., Lueb R., Pollock W., Seiler W., Wartburg A. and Zimmermann P. (1985) Tropospheric chemical composi-

tion measurements in Brazil during the drq season. .I utmos. Chem. 2, 233-256. Delany A. C., Haagensen P., Walters S., Wartburg A. 1,. and Crutzen P. J. (1985) Photochemically produced ozone in the emission from large scale tropical vegetation fires. .I gt,ophps. Res. 90, 2425-2429. Fraser P. J., Hyson P., Rasmussen R. A., Crawford A. J. and Khalil M. A. K. (1986) Methane, carbon monoxide and methylchloroform in the Southern Hemisphere. JI. ~~rmo.s. Chem. 4, 342. Greenberg J. P., Zimmerman P. R., Heidt L. and Pollock W. (I 984) Hydrocarbon and carbon monoxide emissions from biomass burning in Brazil. J. geophys. Res. 89, 1350--1354. Heidt L. E., Krasnec J. P., Lueb R. A.. Pollock W. H., Henry B. E. and Crutzen P. J. (1980) Latitudinal distributions of CO and CH, over the Pacific. J. geophys. Res. 85, 7329-7336. Khalil M. A. K. and Rasmussen R. A. (1985) Causes of increasing atmospheric methane: depletion of hydroxyl radicals and the rise of emissions. Atmospheric Ennironment 19, 397407. KirchholT V. W. J. H. (1984) Are Northern Hemisphere tropospheric ozone densities larger? Eos Trans. AGU 65, 449. Kirchhoff V. W. J. H., Hilsenrath E., Motta A. G., Sahai Y. and Medrano-B R. A. (1983) Equatorial ozone characteristics as measured at Natal (5.9”S, 35.2”W). J. geophys. Res. 88, 6812-6818. Kirchhoff V. W. J. H., Sahai Y. and Motta A. G. (1981) First ozone profiles measured with ECC sondes at Natal (5.9%. 35.2%). Geophys. Res. Lett. 8, 1171-1172. Logan J. A. and Kirchhoff V. W. J. H. (1986) Seasonal variations of tropospheric ozone at Natal, Brazil. J. geophys. Res. 91,7875-7881. Rinsland C. P. and Levine J. S. (1985) Free trouospheric carbon monoxide concentrations in 1’950 and 1951 ‘deduced from infrared total column amount measurements. Nature 318, 25&254. Seiler W. (1974) The cycle of atmospheric CO. TV//US 2, 116-135. Seiler W., Giehl H., Brunke E. G. and Halliday E. (1984) The seasonality of CO abundance in the Southern Hemisphere. Tellus 36, 219-231. Wofsy S. (1989) Amazonian atmospheric studies. Preliminary report. (in preparation).