- Email: [email protected]
The anthropogenic contribution to isoprene concentrations in a rural atmosphere
Atmospheric Environment 34 (2000) 109}115
The anthropogenic contribution to isoprene concentrations in a rural atmosphere Stefan Reimann!,*, Pierluigi Calanca", Peter Hofer! !Swiss Federal Laboratories for Materials Testing and Research (EMPA), Ueberlandstr. 129, CH-8600 Du( bendorf, Switzerland "Swiss Federal Institute of Technology (ETH), Winterthurerstr. 190, CH-8057 Zu( rich, Switzerland Received 16 July 1998; received in revised form 27 May 1999; accepted 8 June 1999
Abstract Atmospheric hydrocarbons are continuously monitored at the rural site of Taenikon, Switzerland. As expected for a rural area, highest isoprene concentrations are found in summer. However, elevated concentrations are also measured on some occasions in winter, in particular during events with long-lasting surface inversions, temperatures constantly below 03C and snow covering the vegetation. During such events, concentrations of isoprene are strongly correlated with those of 1,3-butadiene, a substance that is mainly due to human activities. For these periods, a molar ratio between the concentrations of isoprene and those of 1,3-butadiene of 0.42 is observed. This value, together with the concentrations of 1,3-butadiene, is used to estimate the anthropogenic fraction of the atmospheric isoprene for the whole of 1997. It is found that the fraction is close to 100% in January}February and again in November}December. On the other hand, as early as March, a considerable amount of the observed isoprene appears to be of biogenic origin, although isoprene emissions by trees are negligible. The relative anthropogenic contribution is minimal in midsummer, when biogenic emissions are highest. For this time of the year, the anthropogenic contribution is largest during the early morning hours, in agreement with the tra$c peak on nearby country roads. ( 1999 Elsevier Science Ltd. All rights reserved. Keywords: Non-methane hydrocarbons; Isoprene, 1,3-butadiene; Motor vehicle exhaust
1. Introduction Isoprene is naturally emitted into the atmosphere by plants. In Switzerland, the oak is the most important emitting species, followed by broad-leaved deciduous trees and the Norway spruce (Andreani-Aksoyoglu and Keller, 1995; Simpson, et al., 1999). Inside the leaves, the production of isoprene is driven by photosynthesis and depends on both temperature and solar radiation (Guenther et al., 1993). Therefore, biogenic isoprene emissions are highest during warm summer days. The in#uence of radiation and temperature on isoprene emissions can be described de"ning a biogenic emission potential. This can be calculated using the for-
* Corresponding author. Fax: #41-1821-6244. E-mail address: [email protected] (S. Reimann)
mulation proposed by Guenther et al. (1993, Eqs. (2) and (3)). Recent studies have shown that isoprene correlates with anthropogenic substances both in rural winter atmospheres (Burgess and Penkett, 1993) as well as in urban environments (McLaren et al., 1996; Derwent et al., 1995). This raises the question whether a signi"cant fraction of isoprene can be attributed to human activities. This hypothesis seems to be supported by the measurements of isoprene and 22 other hydrocarbons carried out by the Swiss Federal Laboratories for Materials Testing and Research (EMPA) at Taenikon, a rural station in north-eastern Switzerland. In fact, on occasions, concentrations of isoprene in winter can be as high as in summer (Fig. 1), although the biogenic emission potential is practically zero during the cold season. The present work is an attempt to assess the anthropogenic contribution to measured isoprene concentrations
1352-2310/99/$ - see front matter ( 1999 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 2 - 2 3 1 0 ( 9 9 ) 0 0 2 8 5 - X
110
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
Fig. 1. Annual variations of the measured isoprene concentrations (bold line) and of the biogenic isoprene emission potential (thin line) for 1997. The biogenic isoprene emission potential, which accounts for the temperature and radiation factors, was calculated according to Guenther et al. (1993). Curves are smoothed with a 7-days running "lter.
described in detail in Solberg et al. (1996). In short, every 2 h a volume of 600 ml (30 min with 20 ml min~1) of ambient air is drawn through a cryogenic trap (!443C). Subsequently, hydrocarbons are removed by thermal desorption (1453C) and are injected directly onto the analytical column. The chemical analysis is performed by gas chromatography with #ame ionisation detection (GCFID, Varian 3400), using a PLOT-column (Al O /KCl, 2 3 50 m]0.53 mm, Chrompack) for the separation of the individual hydrocarbons. No ozone scrubber is used to prevent isoprene from decaying on the trap. As it turned out in earlier experiments, the scrubber (Ascarite (II), Aldrich) has no detectable impact on isoprene concentrations even in the presence of signi"cant amounts of ozone. The meteorological data are obtained from the Swiss Meteorological Institute (SMA), which operates a station within the Swiss Automated Observational Network (ANETZ) at the same site.
3. Estimating the anthropogenic contribution to atmospheric isoprene in a rural atmosphere using the data collected at Taenikon. We start by assuming that concentrations of anthropogenic isoprene are proportional to those of 1,3butadiene at any time of the year. Thereby, the constant of proportionality is determined from winter samples una!ected by biogenic emissions. We then proceed by computing the anthropogenic contribution to isoprene as the ratio between the calculated concentrations of anthropogenic isoprene and the observed concentrations. For this latter quantity, we "nally discuss the annual and diurnal cycles.
2. Measuring site and data acquisition The rural site of Taenikon (coord.: 473 29@ N, 83 54@ E, 540 m a.s.l.) is part of the EMEP-network (co-operative programme for monitoring and evaluation of the longrange transmission of air pollutants in Europe). It is situated in the north-eastern part of Switzerland in a topographic depression surrounded by a rugged landscape, with mixed forests on the hills and farming in the gentle valleys. Within EMEP, Taenikon is classi"ed as a rural station, in spite of several anthropogenic sources of pollutants at the local and regional scale: roads of regional interest and agriculture in the immediate vicinity, the village of Aadorf (about 8000 inhabitants) 1.5 km to the north, a motorway approximately 5 km to the east and the agglomeration of Zurich about 40 km to the south-west. The quasi-continuous measurements of isoprene and 22 other hydrocarbons from ethane up to o-xylene are
As mentioned in the introduction, we rely on 1,3butadiene to identify anthropogenic isoprene. 1,3-butadiene is well suited as a tracer because in urban environment it is predominantly emitted by motor vehicles (Ye et al., 1997; Derwent et al., 1999). The chemical properties of 1,3-butadiene are also of relevance, as this organic compound possesses a structure similar to that of isoprene and is characterised by a comparable rate regarding the reaction with OH (though the di!erence between the rate coe$cients is signi"cant during summer, see discussion below). In a "rst step, we determine the molar ratio of isoprene to 1,3-butadiene for two winter events, the "rst in December 1996, the second in January 1997, characterised by temperatures continuously below the freezing point, strong surface inversions and snow covering the vegetation. Under these conditions, the biogenic emission potential is zero (Fig. 1). For this reason the molar ratio between the observed isoprene concentrations and those of 1,3-butadiene can be interpreted as the ratio between anthropogenic isoprene and 1,3-butadiene. In a second step, we assume that the molar ratio between anthropogenic isoprene and 1,3-butadiene at the measuring site does not vary in time. We then use the average of the values for the two events in December 1996 and January 1997 to estimate the concentrations of anthropogenic isoprene for the whole of 1997. Finally, we de"ne the anthropogenic contribution as the ratio between calculated anthropogenic and measured total isoprene. The time series of isoprene and 1,3-butadiene for December 1996 and January 1997 are plotted in Fig. 2.
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
111
Fig. 2. Isoprene and 1,3-butadiene concentrations at Taenikon for two events in December 1996 (a) and January 1997 (b).
Fig. 3. Scatterplot of isoprene versus 1,3-butadiene for the data of Fig. 2. The regression line forced through zero is shown as a continuous line, whereas the free regression line is displayed as a dashed line. The regression equations and the corresponding correlation coe$cients are also indicated.
The parallelism is evident. As shown by Fig. 3, the relation between isoprene and 1,3-butadiene is indeed nearly linear. Regression analysis with the regression constant forced to zero, yields a regression coe$cient of 0.415$0.003 and correlation coe$cient squared of 0.940, whereas allowing for an intercept yields a regression coe$cient of 0.444$0.007, a regression constant of !0.009$0.003 and a correlation coe$cient squared of 0.945. With respect to the standard errors in the regression parameters, the di!erence between the two sets appears to be insigni"cant. However, there are reasons for forcing the regression line through the origin. In "rst place, this is
Fig. 4. Annual variations of the calculated anthropogenic isoprene concentrations. The values calculated with the coe$cient of the forced regression of Fig. 4 are shown as a thick line, those derived using the coe$cient and intercept of the free regression of Fig. 4 are displayed as thin line. Curves are smoothed with a 7-days running "lter.
consistent with assuming the same sources for the two substances. Secondly, as the concentrations of 1,3-butadiene are very small in summer, even the small negative value for the intercept of !0.009$0.003 makes most of the calculated anthropogenic isoprene concentrations negative during this season (Fig. 4). The numeric value of 0.42 for the coe$cient of the forced regression is comparable to values of the molar ratio between isoprene and 1,3-butadiene obtained in recent studies: 0.38 for a suburban area in Toronto during winter (McLaren et al., 1996), 0.35}0.39 for an urban location within the city of London (Derwent et al., 1995) and 0.33}0.50 for a rural atmosphere in England during winter (Burgess and Penkett, 1993). Nonetheless, our
112
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
Table 1 Reaction rate constants from Atkinson (1994) for the reactions of isoprene and 1,3-butadiene with hydroxyl-radicals (OH) resp. ozone (O ) 3 k[cm3 mol~1 s~1]
OH
O 3
isoprene 1,3-butadiene
101.0]10~12 66.6]10~12
12.8]10~18 6.3]10~18
assumption of a constant molar ratio remains arguable, since samples taken at di!erent times of the day and of the year represent air plumes with di!erent origins and histories. In particular, during summer the relative abundance of any two non-methane hydrocarbons is strongly dependent on their respective reaction rate with OH and (in the case of alkenes) with ozone. The importance of the OH-photochemistry during summer can be shown with simple arguments. Assuming for instance average abundances of the OH-radical of the order of 1]106 molec cm~3 and rate coe$cients as given by Atkinson (1994) (Table 1), the reaction between isoprene and OH is about 1.3 times faster than the reaction between 1,3-butadiene and OH. For equivalent source distribution, transport time and atmospheric mixing as in winter, this leads to an overestimation of the anthropogenic contribution to isoprene by about 30%. During summer afternoons, the situation is even more critical, as OH-abundances can be of the order of 1]107 molec cm~3. In this case the anthropogenic contribution is overestimated by a factor of 10. On the other hand, the ozone photochemistry can be neglected because the reactions of ozone with isoprene and 1,3-butadiene proceed at considerably lower rates than the respective reactions with OH. Even in the extreme case of OH-abundances of the order of 1]107 molec cm~3 and ozone concentrations of the order of and 2.4]1012 molec cm~3 (BUWAL, 1998), the error does not exceed 10%.
4. Annual and daily cycles of anthropogenic and biogenic isoprene As a starting point for the discussion, Fig. 5a depicts the annual courses of the total and of the estimated anthropogenic isoprene concentrations. In addition, the annual course of the anthropogenic contribution is shown in Fig. 5b. In both "gures curves are smoothed by applying a running mean with a 7-day window. The course of the measured isoprene concentrations in Fig. 5a follows quite closely that of the biogenic emission potential in Fig. 1. Highest concentrations are measured during the summer season, when the radiation and tem-
Fig. 5. Annual variation of the anthropogenic and the observed total isoprene (Fig. 5a) and of the anthropogenic fraction of isoprene (Fig. 5b). Curves are smoothed with a 7-days running "lter.
perature conditions are most favourable for biogenic emissions. In contrast, the anthropogenic concentrations are highest in winter. Two explanations can be given for this winter maximum. Firstly, during winter, photochemical reactions are slow and turbulent mixing is modest. Secondly, emissions from motor vehicles are higher as a consequence of cold starts (Heeb et al., 1999). Combining the curves of the anthropogenic and total isoprene concentrations, it is readily shown that the anthropogenic fraction is also highest in winter (Fig. 5b). In fact, it is close to 100% in January and February, and again in November and December. In March, however, it is signi"cantly less than 100%, in spite of temperatures often falling below 03C, particularly during the nights. This drop manifests itself during the afternoon hours (Fig. 6) and indicates the presence of non-negligible amounts of biogenic isoprene. The sources of biogenic isoprene in March are, however, di$cult to single out.
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
Fig. 6. Mean diurnal cycles of the anthropogenic fraction of isoprene in January, February and March 1997.
Obviously, broad-leaved deciduous trees do not enter into consideration as they are still bare at this time of the year. We also tend to exclude conifers, because they start to produce isoprene not earlier than two weeks after the last frost (Steinbrecher, 1998). In 1997, according to the data of the Swiss Meteorological Institute, frost conditions lasted until the end of March. Therefore, isoprene emissions from conifers should not have begun before April. As an alternative, we are inclined to consider moss or lichens as the source of biogenic isoprene in March. We are not in the position to quantify these emissions for our study area, but we would like to point out that, according to a recent study by Janson and De Serves (1998), sphagnum moss is an important source of isoprene in Northern Europe. At our site, a further possibility for the existence of biogenic isoprene in winter and early spring is the advection of air masses from warmer areas in Southern Europe, where evergreen plants emit isoprene during the whole year. However, vigorous advection of air masses over the Alps is mainly limited to Foehn events and these occur too rarely to signi"cantly in#uence isoprene concentrations for time periods of more than a few days. In summer, the anthropogenic fraction typically varies between 10 and 15% (Fig. 5b). As these values are systematically overestimated, (see Section 3) a more cautious assessment for the average fraction in summer is probably of the order of 5%. For a detailed analysis of the summer values, normalised daily cycles of anthropogenic and biogenic isoprene are shown in Fig. 7. The course of the anthropogenic isoprene closely follows that of benzene, an organic compound known to be mainly emitted by motor vehicles. For both substances, highest values are found in the early morning, (6 a.m.) and in the
113
Fig. 7. Normalised diurnal cycles of the anthropogenic and the observed total isoprene concentrations and of the measured benzene concentrations, together with the biogenic isoprene emission potential. Curves are for August 1997.
late afternoon (8 p.m.), approximately at the time when the in#uence of tra$c from nearby roads is maximal. In this sense, the anthropogenic isoprene can also be considered as a product of motor vehicles. In contrast, total isoprene is signi"cantly a!ected by biogenic emissions. Following the rise in the biogenic emission potential, it increases until 8 a.m. (Fig. 7). Thereafter, atmospheric mixing and photochemistry becomes so e!ective that isoprene concentrations tend to remain at a more or less constant level, although the biogenic emission potential becomes largest only around midday. The largest concentrations of total isoprene are not observed until sunset. This evening peak is the consequence of less favourable atmospheric dispersion conditions and smaller amounts of OH and ozone. Moreover, isoprene concentrations at our site are also a!ected by a systematic change in the wind direction from NE to SW (Fig. 8). In the afternoon, winds usually blow from NE and concentrations are small because pasture is the dominant type of surface cover upwind of the station. In the evening, the wind direction changes to SW, and the air sampled at the station has higher concentrations of isoprene, as it has moved above mixed forests in the south. With respect to the anthropogenic isoprene, it is interesting to point out that the molar ratio between isoprene and 1,3-butadiene calculated with the data collected at 6 a.m. in summer does not considerably di!er from the ratio calculated for the winter season. Excluding all cases characterised by wind speeds higher than 2 m s~1, which we assume to correspond to air masses not strictly representative of local conditions, a value of 0.49 is found by regression analysis (Fig. 9). We see this result as an
114
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
Fig. 8. Frequency distribution of the wind direction for August 1997: 12 a.m.}2 p.m. (a); 6 p.m.}8 p.m. (b). Wind speed classes are indicated on the right-hand side.
Fig. 9. Scatterplot of isoprene versus 1,3-butadiene. Data represent samples taken at 6 a.m., July}August 1997. The linear regression was calculated excluding all cases for which the wind speed was larger than 2 m s~1.
indication that the sources of anthropogenic isoprene and 1,3-butadiene in summer are most likely identical to those in winter. Taking into account that anthropogenic isoprene is also correlated to benzene, we conclude that motor vehicles must contribute substantially to anthropogenic isoprene in all seasons.
5. Summary In winter, isoprene concentrations measured at Taenikon (eastern Switzerland) can be as high as in
summer, indicating that the anthropogenic contribution to isoprene is signi"cant even at this rural location. In this study we have therefore attempted to estimate this contribution. The method proposed uses 1,3-butadiene to trace anthropogenic isoprene. For the whole of 1997 the anthropogenic isoprene does not contribute more than 30% to the observed total concentrations. The anthropogenic contribution is close to 100% in January and February, and again in November and December. As early as March, however, the anthropogenic contribution signi"cantly drops below 100%, a fact which we explain by the appearance of biogenic isoprene during the afternoon hours. The anthropogenic contribution is minimal in summer, with estimated values of 5}10%. Nevertheless, higher values are calculated for the early morning hours. These are well timed with the peak in the concentrations of benzene. Since motor vehicles are a major source of benzene and since tra$c on regional roads and highways in the surroundings of our station is also largest at about the same time of the day, we propose that motor vehicles are responsible for a signi"cant portion of the anthropogenic isoprene, a thesis already put forward by Derwent et al. (1995) and McLaren et al. (1996). The attribution of anthropogenic isoprene to tra$c emissions would explain why values of the molar ratio between total isoprene and 1,3-butadiene during earlymorning hours in summer are comparable to those calculated for the winter season, during episodes characterised by strong surface inversions. Furthermore, this would also indicate that the composition of fuel does not change considerably during the course of the year. Only a few studies have attempted to measure isoprene and other organic compounds directly in car exhausts
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
(e.g. CARB, 1993; Jemma et al., 1995). In general, ratios of isoprene to 1,3-butadiene based on these studies vary considerably and depend on the type of motors and fuels. Basically, direct measurements of anthropogenic isoprene would be more accurate than our approximate procedure. However, unless very speci"c emission inventories are available, it is di$cult to use the results of such studies for assessing the anthropogenic contribution to isoprene concentrations at a given site. For the time being, the approach adopted in the present work should, therefore, represent a valid alternative.
Acknowledgements The measurements at Taenikon were sponsored by the Swiss Federal O$ce of Environment, Forests and Landscape (BUWAL). Meteorological data were kindly provided by the Swiss Meteorological Institute (SMA) and prepared by J. Forrer (EMPA). We are grateful to R. Gehrig (EMPA), D. Simpson (EMEP-MSC-W, DNMI, Oslo) and R. Steinbrecher (IfU, Garmisch-Partenkirchen) for helpful discussions and to two anonymous reviewers for useful comments.
References Andreani-Aksoyoglu, S., Keller, J., 1995. Estimates of monoterpene and isoprene emissions from the forests of Switzerland. Journal of Atmospheric Chemistry 20, 71}87. Atkinson, R., 1994. Gas-phase tropospheric chemistry of organic compounds. Journal of Physical and Chemical Reference Data. Monograph No. 2, 1}216. Burgess, R.A., Penkett, S.A., 1993. Ground-based non-methane hydrocarbon measurements in England. In: Borrell, P. (Ed.), Proceedings of the EUROTRAC '92, Academic Publishing, The Hague, The Netherlands, pp. 165}169. BUWAL, 1998. Messresultate des Nationalen Beobachtungsnetzes fuK r Luftfremdsto!e (NABEL). Schriftenreihe Umwelt Nr. 303, Bern. CARB, 1993. Establishment of corrections to reactivity adjustment factors for transitional low-emission vehicles and low-
115
emission vehicles operating on phase 2 reformulated gasoline, airshed modelling protocol and "nal results. California Air Resources Board (CARB). Derwent, R.G., Middleton, D.R., Field, R.A., Goldstone, M.E., Lester, J.N., Perry, R., 1995. Analysis and interpretation of air quality data from an urban roadside location in central London over the period from July 1991 to July 1992. Atmospheric Environment 29 (8), 923}946. Derwent, R.G., Field, R.A., Dollard, G.J., Davies, T.J., Dumitrean, P., Pepler, S.A., 1999. Analysis and interpretation of the continuous hourly monitoring data for 26 C2}C8 Hydrocarbons at twelve United Kingdom sites during 1996. Atmospheric Environment, (in press). Guenther, A.B., Zimmerman, P.R., Harley, P.C., 1993. Isoprene and monoterpene emission rate variability: model evaluations and sensitivity analyses. Journal of Geophysical Research 99 (D7), 12609}12617. Heeb, N.V., Forss, A.-F., Bach, C., 1999. Fast and quantitative measurements of benzene, toluene and C2-benzenes in automotive exhaust during transient engine operation with and without catalytic exhaust gas treatment. Atmospheric Environment 33, 205}215. Janson, R., DeServes, C., 1998. Isoprene emissions form boreal wetlands in Scandinavia. Journal of Geophysical Research 103 (D19), 25513}25519. Jemma, C.A., Shore, P.R., Widdicombe, K.A., 1995. Analysis of C }C hydrocarbons using dual-column capillary GC: ap1 16 plication to exhaust emission from passenger car and motorcycle engines. Journal of Chromatographic Science 33, 34}48. McLaren, R., Singleton, D.L., Lai, J.Y.K., Khouw, B., Singer, E., Wu, Z., Niki, H., 1996. Analysis of motor vehicle sources and their contribution to ambient hydrocarbon distributions at urban sites in Toronto during the Southern oxidants study. Atmospheric Environment 30 (12), 2219}2232. Simpson, D., et al., 1999. Inventorying emissions from nature in Europe. Journal of Geophysical Research 104 (7), 8113}8152. Solberg, S., Dye, C., Schmidbauer, N., Herzog, A., Gehrig, R., 1996. Carbonyls and non-methane hydrocarbons at rural sites from the Mediterranean to the Arctic. Journal of Atmospheric Chemistry 25, 33}66. Steinbrecher, R., 1998. personal communication. Ye, Y., Galbally, I.E., Weeks, I.A., 1997. Emission of 1,3-Butadiene from petrol-driven motor vehicles. Atmospheric Environment 31 (8), 1157}1165.
The anthropogenic contribution to isoprene concentrations in a rural atmosphere Stefan Reimann!,*, Pierluigi Calanca", Peter Hofer! !Swiss Federal Laboratories for Materials Testing and Research (EMPA), Ueberlandstr. 129, CH-8600 Du( bendorf, Switzerland "Swiss Federal Institute of Technology (ETH), Winterthurerstr. 190, CH-8057 Zu( rich, Switzerland Received 16 July 1998; received in revised form 27 May 1999; accepted 8 June 1999
Abstract Atmospheric hydrocarbons are continuously monitored at the rural site of Taenikon, Switzerland. As expected for a rural area, highest isoprene concentrations are found in summer. However, elevated concentrations are also measured on some occasions in winter, in particular during events with long-lasting surface inversions, temperatures constantly below 03C and snow covering the vegetation. During such events, concentrations of isoprene are strongly correlated with those of 1,3-butadiene, a substance that is mainly due to human activities. For these periods, a molar ratio between the concentrations of isoprene and those of 1,3-butadiene of 0.42 is observed. This value, together with the concentrations of 1,3-butadiene, is used to estimate the anthropogenic fraction of the atmospheric isoprene for the whole of 1997. It is found that the fraction is close to 100% in January}February and again in November}December. On the other hand, as early as March, a considerable amount of the observed isoprene appears to be of biogenic origin, although isoprene emissions by trees are negligible. The relative anthropogenic contribution is minimal in midsummer, when biogenic emissions are highest. For this time of the year, the anthropogenic contribution is largest during the early morning hours, in agreement with the tra$c peak on nearby country roads. ( 1999 Elsevier Science Ltd. All rights reserved. Keywords: Non-methane hydrocarbons; Isoprene, 1,3-butadiene; Motor vehicle exhaust
1. Introduction Isoprene is naturally emitted into the atmosphere by plants. In Switzerland, the oak is the most important emitting species, followed by broad-leaved deciduous trees and the Norway spruce (Andreani-Aksoyoglu and Keller, 1995; Simpson, et al., 1999). Inside the leaves, the production of isoprene is driven by photosynthesis and depends on both temperature and solar radiation (Guenther et al., 1993). Therefore, biogenic isoprene emissions are highest during warm summer days. The in#uence of radiation and temperature on isoprene emissions can be described de"ning a biogenic emission potential. This can be calculated using the for-
* Corresponding author. Fax: #41-1821-6244. E-mail address: [email protected] (S. Reimann)
mulation proposed by Guenther et al. (1993, Eqs. (2) and (3)). Recent studies have shown that isoprene correlates with anthropogenic substances both in rural winter atmospheres (Burgess and Penkett, 1993) as well as in urban environments (McLaren et al., 1996; Derwent et al., 1995). This raises the question whether a signi"cant fraction of isoprene can be attributed to human activities. This hypothesis seems to be supported by the measurements of isoprene and 22 other hydrocarbons carried out by the Swiss Federal Laboratories for Materials Testing and Research (EMPA) at Taenikon, a rural station in north-eastern Switzerland. In fact, on occasions, concentrations of isoprene in winter can be as high as in summer (Fig. 1), although the biogenic emission potential is practically zero during the cold season. The present work is an attempt to assess the anthropogenic contribution to measured isoprene concentrations
1352-2310/99/$ - see front matter ( 1999 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 2 - 2 3 1 0 ( 9 9 ) 0 0 2 8 5 - X
110
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
Fig. 1. Annual variations of the measured isoprene concentrations (bold line) and of the biogenic isoprene emission potential (thin line) for 1997. The biogenic isoprene emission potential, which accounts for the temperature and radiation factors, was calculated according to Guenther et al. (1993). Curves are smoothed with a 7-days running "lter.
described in detail in Solberg et al. (1996). In short, every 2 h a volume of 600 ml (30 min with 20 ml min~1) of ambient air is drawn through a cryogenic trap (!443C). Subsequently, hydrocarbons are removed by thermal desorption (1453C) and are injected directly onto the analytical column. The chemical analysis is performed by gas chromatography with #ame ionisation detection (GCFID, Varian 3400), using a PLOT-column (Al O /KCl, 2 3 50 m]0.53 mm, Chrompack) for the separation of the individual hydrocarbons. No ozone scrubber is used to prevent isoprene from decaying on the trap. As it turned out in earlier experiments, the scrubber (Ascarite (II), Aldrich) has no detectable impact on isoprene concentrations even in the presence of signi"cant amounts of ozone. The meteorological data are obtained from the Swiss Meteorological Institute (SMA), which operates a station within the Swiss Automated Observational Network (ANETZ) at the same site.
3. Estimating the anthropogenic contribution to atmospheric isoprene in a rural atmosphere using the data collected at Taenikon. We start by assuming that concentrations of anthropogenic isoprene are proportional to those of 1,3butadiene at any time of the year. Thereby, the constant of proportionality is determined from winter samples una!ected by biogenic emissions. We then proceed by computing the anthropogenic contribution to isoprene as the ratio between the calculated concentrations of anthropogenic isoprene and the observed concentrations. For this latter quantity, we "nally discuss the annual and diurnal cycles.
2. Measuring site and data acquisition The rural site of Taenikon (coord.: 473 29@ N, 83 54@ E, 540 m a.s.l.) is part of the EMEP-network (co-operative programme for monitoring and evaluation of the longrange transmission of air pollutants in Europe). It is situated in the north-eastern part of Switzerland in a topographic depression surrounded by a rugged landscape, with mixed forests on the hills and farming in the gentle valleys. Within EMEP, Taenikon is classi"ed as a rural station, in spite of several anthropogenic sources of pollutants at the local and regional scale: roads of regional interest and agriculture in the immediate vicinity, the village of Aadorf (about 8000 inhabitants) 1.5 km to the north, a motorway approximately 5 km to the east and the agglomeration of Zurich about 40 km to the south-west. The quasi-continuous measurements of isoprene and 22 other hydrocarbons from ethane up to o-xylene are
As mentioned in the introduction, we rely on 1,3butadiene to identify anthropogenic isoprene. 1,3-butadiene is well suited as a tracer because in urban environment it is predominantly emitted by motor vehicles (Ye et al., 1997; Derwent et al., 1999). The chemical properties of 1,3-butadiene are also of relevance, as this organic compound possesses a structure similar to that of isoprene and is characterised by a comparable rate regarding the reaction with OH (though the di!erence between the rate coe$cients is signi"cant during summer, see discussion below). In a "rst step, we determine the molar ratio of isoprene to 1,3-butadiene for two winter events, the "rst in December 1996, the second in January 1997, characterised by temperatures continuously below the freezing point, strong surface inversions and snow covering the vegetation. Under these conditions, the biogenic emission potential is zero (Fig. 1). For this reason the molar ratio between the observed isoprene concentrations and those of 1,3-butadiene can be interpreted as the ratio between anthropogenic isoprene and 1,3-butadiene. In a second step, we assume that the molar ratio between anthropogenic isoprene and 1,3-butadiene at the measuring site does not vary in time. We then use the average of the values for the two events in December 1996 and January 1997 to estimate the concentrations of anthropogenic isoprene for the whole of 1997. Finally, we de"ne the anthropogenic contribution as the ratio between calculated anthropogenic and measured total isoprene. The time series of isoprene and 1,3-butadiene for December 1996 and January 1997 are plotted in Fig. 2.
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
111
Fig. 2. Isoprene and 1,3-butadiene concentrations at Taenikon for two events in December 1996 (a) and January 1997 (b).
Fig. 3. Scatterplot of isoprene versus 1,3-butadiene for the data of Fig. 2. The regression line forced through zero is shown as a continuous line, whereas the free regression line is displayed as a dashed line. The regression equations and the corresponding correlation coe$cients are also indicated.
The parallelism is evident. As shown by Fig. 3, the relation between isoprene and 1,3-butadiene is indeed nearly linear. Regression analysis with the regression constant forced to zero, yields a regression coe$cient of 0.415$0.003 and correlation coe$cient squared of 0.940, whereas allowing for an intercept yields a regression coe$cient of 0.444$0.007, a regression constant of !0.009$0.003 and a correlation coe$cient squared of 0.945. With respect to the standard errors in the regression parameters, the di!erence between the two sets appears to be insigni"cant. However, there are reasons for forcing the regression line through the origin. In "rst place, this is
Fig. 4. Annual variations of the calculated anthropogenic isoprene concentrations. The values calculated with the coe$cient of the forced regression of Fig. 4 are shown as a thick line, those derived using the coe$cient and intercept of the free regression of Fig. 4 are displayed as thin line. Curves are smoothed with a 7-days running "lter.
consistent with assuming the same sources for the two substances. Secondly, as the concentrations of 1,3-butadiene are very small in summer, even the small negative value for the intercept of !0.009$0.003 makes most of the calculated anthropogenic isoprene concentrations negative during this season (Fig. 4). The numeric value of 0.42 for the coe$cient of the forced regression is comparable to values of the molar ratio between isoprene and 1,3-butadiene obtained in recent studies: 0.38 for a suburban area in Toronto during winter (McLaren et al., 1996), 0.35}0.39 for an urban location within the city of London (Derwent et al., 1995) and 0.33}0.50 for a rural atmosphere in England during winter (Burgess and Penkett, 1993). Nonetheless, our
112
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
Table 1 Reaction rate constants from Atkinson (1994) for the reactions of isoprene and 1,3-butadiene with hydroxyl-radicals (OH) resp. ozone (O ) 3 k[cm3 mol~1 s~1]
OH
O 3
isoprene 1,3-butadiene
101.0]10~12 66.6]10~12
12.8]10~18 6.3]10~18
assumption of a constant molar ratio remains arguable, since samples taken at di!erent times of the day and of the year represent air plumes with di!erent origins and histories. In particular, during summer the relative abundance of any two non-methane hydrocarbons is strongly dependent on their respective reaction rate with OH and (in the case of alkenes) with ozone. The importance of the OH-photochemistry during summer can be shown with simple arguments. Assuming for instance average abundances of the OH-radical of the order of 1]106 molec cm~3 and rate coe$cients as given by Atkinson (1994) (Table 1), the reaction between isoprene and OH is about 1.3 times faster than the reaction between 1,3-butadiene and OH. For equivalent source distribution, transport time and atmospheric mixing as in winter, this leads to an overestimation of the anthropogenic contribution to isoprene by about 30%. During summer afternoons, the situation is even more critical, as OH-abundances can be of the order of 1]107 molec cm~3. In this case the anthropogenic contribution is overestimated by a factor of 10. On the other hand, the ozone photochemistry can be neglected because the reactions of ozone with isoprene and 1,3-butadiene proceed at considerably lower rates than the respective reactions with OH. Even in the extreme case of OH-abundances of the order of 1]107 molec cm~3 and ozone concentrations of the order of and 2.4]1012 molec cm~3 (BUWAL, 1998), the error does not exceed 10%.
4. Annual and daily cycles of anthropogenic and biogenic isoprene As a starting point for the discussion, Fig. 5a depicts the annual courses of the total and of the estimated anthropogenic isoprene concentrations. In addition, the annual course of the anthropogenic contribution is shown in Fig. 5b. In both "gures curves are smoothed by applying a running mean with a 7-day window. The course of the measured isoprene concentrations in Fig. 5a follows quite closely that of the biogenic emission potential in Fig. 1. Highest concentrations are measured during the summer season, when the radiation and tem-
Fig. 5. Annual variation of the anthropogenic and the observed total isoprene (Fig. 5a) and of the anthropogenic fraction of isoprene (Fig. 5b). Curves are smoothed with a 7-days running "lter.
perature conditions are most favourable for biogenic emissions. In contrast, the anthropogenic concentrations are highest in winter. Two explanations can be given for this winter maximum. Firstly, during winter, photochemical reactions are slow and turbulent mixing is modest. Secondly, emissions from motor vehicles are higher as a consequence of cold starts (Heeb et al., 1999). Combining the curves of the anthropogenic and total isoprene concentrations, it is readily shown that the anthropogenic fraction is also highest in winter (Fig. 5b). In fact, it is close to 100% in January and February, and again in November and December. In March, however, it is signi"cantly less than 100%, in spite of temperatures often falling below 03C, particularly during the nights. This drop manifests itself during the afternoon hours (Fig. 6) and indicates the presence of non-negligible amounts of biogenic isoprene. The sources of biogenic isoprene in March are, however, di$cult to single out.
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
Fig. 6. Mean diurnal cycles of the anthropogenic fraction of isoprene in January, February and March 1997.
Obviously, broad-leaved deciduous trees do not enter into consideration as they are still bare at this time of the year. We also tend to exclude conifers, because they start to produce isoprene not earlier than two weeks after the last frost (Steinbrecher, 1998). In 1997, according to the data of the Swiss Meteorological Institute, frost conditions lasted until the end of March. Therefore, isoprene emissions from conifers should not have begun before April. As an alternative, we are inclined to consider moss or lichens as the source of biogenic isoprene in March. We are not in the position to quantify these emissions for our study area, but we would like to point out that, according to a recent study by Janson and De Serves (1998), sphagnum moss is an important source of isoprene in Northern Europe. At our site, a further possibility for the existence of biogenic isoprene in winter and early spring is the advection of air masses from warmer areas in Southern Europe, where evergreen plants emit isoprene during the whole year. However, vigorous advection of air masses over the Alps is mainly limited to Foehn events and these occur too rarely to signi"cantly in#uence isoprene concentrations for time periods of more than a few days. In summer, the anthropogenic fraction typically varies between 10 and 15% (Fig. 5b). As these values are systematically overestimated, (see Section 3) a more cautious assessment for the average fraction in summer is probably of the order of 5%. For a detailed analysis of the summer values, normalised daily cycles of anthropogenic and biogenic isoprene are shown in Fig. 7. The course of the anthropogenic isoprene closely follows that of benzene, an organic compound known to be mainly emitted by motor vehicles. For both substances, highest values are found in the early morning, (6 a.m.) and in the
113
Fig. 7. Normalised diurnal cycles of the anthropogenic and the observed total isoprene concentrations and of the measured benzene concentrations, together with the biogenic isoprene emission potential. Curves are for August 1997.
late afternoon (8 p.m.), approximately at the time when the in#uence of tra$c from nearby roads is maximal. In this sense, the anthropogenic isoprene can also be considered as a product of motor vehicles. In contrast, total isoprene is signi"cantly a!ected by biogenic emissions. Following the rise in the biogenic emission potential, it increases until 8 a.m. (Fig. 7). Thereafter, atmospheric mixing and photochemistry becomes so e!ective that isoprene concentrations tend to remain at a more or less constant level, although the biogenic emission potential becomes largest only around midday. The largest concentrations of total isoprene are not observed until sunset. This evening peak is the consequence of less favourable atmospheric dispersion conditions and smaller amounts of OH and ozone. Moreover, isoprene concentrations at our site are also a!ected by a systematic change in the wind direction from NE to SW (Fig. 8). In the afternoon, winds usually blow from NE and concentrations are small because pasture is the dominant type of surface cover upwind of the station. In the evening, the wind direction changes to SW, and the air sampled at the station has higher concentrations of isoprene, as it has moved above mixed forests in the south. With respect to the anthropogenic isoprene, it is interesting to point out that the molar ratio between isoprene and 1,3-butadiene calculated with the data collected at 6 a.m. in summer does not considerably di!er from the ratio calculated for the winter season. Excluding all cases characterised by wind speeds higher than 2 m s~1, which we assume to correspond to air masses not strictly representative of local conditions, a value of 0.49 is found by regression analysis (Fig. 9). We see this result as an
114
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
Fig. 8. Frequency distribution of the wind direction for August 1997: 12 a.m.}2 p.m. (a); 6 p.m.}8 p.m. (b). Wind speed classes are indicated on the right-hand side.
Fig. 9. Scatterplot of isoprene versus 1,3-butadiene. Data represent samples taken at 6 a.m., July}August 1997. The linear regression was calculated excluding all cases for which the wind speed was larger than 2 m s~1.
indication that the sources of anthropogenic isoprene and 1,3-butadiene in summer are most likely identical to those in winter. Taking into account that anthropogenic isoprene is also correlated to benzene, we conclude that motor vehicles must contribute substantially to anthropogenic isoprene in all seasons.
5. Summary In winter, isoprene concentrations measured at Taenikon (eastern Switzerland) can be as high as in
summer, indicating that the anthropogenic contribution to isoprene is signi"cant even at this rural location. In this study we have therefore attempted to estimate this contribution. The method proposed uses 1,3-butadiene to trace anthropogenic isoprene. For the whole of 1997 the anthropogenic isoprene does not contribute more than 30% to the observed total concentrations. The anthropogenic contribution is close to 100% in January and February, and again in November and December. As early as March, however, the anthropogenic contribution signi"cantly drops below 100%, a fact which we explain by the appearance of biogenic isoprene during the afternoon hours. The anthropogenic contribution is minimal in summer, with estimated values of 5}10%. Nevertheless, higher values are calculated for the early morning hours. These are well timed with the peak in the concentrations of benzene. Since motor vehicles are a major source of benzene and since tra$c on regional roads and highways in the surroundings of our station is also largest at about the same time of the day, we propose that motor vehicles are responsible for a signi"cant portion of the anthropogenic isoprene, a thesis already put forward by Derwent et al. (1995) and McLaren et al. (1996). The attribution of anthropogenic isoprene to tra$c emissions would explain why values of the molar ratio between total isoprene and 1,3-butadiene during earlymorning hours in summer are comparable to those calculated for the winter season, during episodes characterised by strong surface inversions. Furthermore, this would also indicate that the composition of fuel does not change considerably during the course of the year. Only a few studies have attempted to measure isoprene and other organic compounds directly in car exhausts
S. Reimann et al. / Atmospheric Environment 34 (2000) 109}115
(e.g. CARB, 1993; Jemma et al., 1995). In general, ratios of isoprene to 1,3-butadiene based on these studies vary considerably and depend on the type of motors and fuels. Basically, direct measurements of anthropogenic isoprene would be more accurate than our approximate procedure. However, unless very speci"c emission inventories are available, it is di$cult to use the results of such studies for assessing the anthropogenic contribution to isoprene concentrations at a given site. For the time being, the approach adopted in the present work should, therefore, represent a valid alternative.
Acknowledgements The measurements at Taenikon were sponsored by the Swiss Federal O$ce of Environment, Forests and Landscape (BUWAL). Meteorological data were kindly provided by the Swiss Meteorological Institute (SMA) and prepared by J. Forrer (EMPA). We are grateful to R. Gehrig (EMPA), D. Simpson (EMEP-MSC-W, DNMI, Oslo) and R. Steinbrecher (IfU, Garmisch-Partenkirchen) for helpful discussions and to two anonymous reviewers for useful comments.
References Andreani-Aksoyoglu, S., Keller, J., 1995. Estimates of monoterpene and isoprene emissions from the forests of Switzerland. Journal of Atmospheric Chemistry 20, 71}87. Atkinson, R., 1994. Gas-phase tropospheric chemistry of organic compounds. Journal of Physical and Chemical Reference Data. Monograph No. 2, 1}216. Burgess, R.A., Penkett, S.A., 1993. Ground-based non-methane hydrocarbon measurements in England. In: Borrell, P. (Ed.), Proceedings of the EUROTRAC '92, Academic Publishing, The Hague, The Netherlands, pp. 165}169. BUWAL, 1998. Messresultate des Nationalen Beobachtungsnetzes fuK r Luftfremdsto!e (NABEL). Schriftenreihe Umwelt Nr. 303, Bern. CARB, 1993. Establishment of corrections to reactivity adjustment factors for transitional low-emission vehicles and low-
115
emission vehicles operating on phase 2 reformulated gasoline, airshed modelling protocol and "nal results. California Air Resources Board (CARB). Derwent, R.G., Middleton, D.R., Field, R.A., Goldstone, M.E., Lester, J.N., Perry, R., 1995. Analysis and interpretation of air quality data from an urban roadside location in central London over the period from July 1991 to July 1992. Atmospheric Environment 29 (8), 923}946. Derwent, R.G., Field, R.A., Dollard, G.J., Davies, T.J., Dumitrean, P., Pepler, S.A., 1999. Analysis and interpretation of the continuous hourly monitoring data for 26 C2}C8 Hydrocarbons at twelve United Kingdom sites during 1996. Atmospheric Environment, (in press). Guenther, A.B., Zimmerman, P.R., Harley, P.C., 1993. Isoprene and monoterpene emission rate variability: model evaluations and sensitivity analyses. Journal of Geophysical Research 99 (D7), 12609}12617. Heeb, N.V., Forss, A.-F., Bach, C., 1999. Fast and quantitative measurements of benzene, toluene and C2-benzenes in automotive exhaust during transient engine operation with and without catalytic exhaust gas treatment. Atmospheric Environment 33, 205}215. Janson, R., DeServes, C., 1998. Isoprene emissions form boreal wetlands in Scandinavia. Journal of Geophysical Research 103 (D19), 25513}25519. Jemma, C.A., Shore, P.R., Widdicombe, K.A., 1995. Analysis of C }C hydrocarbons using dual-column capillary GC: ap1 16 plication to exhaust emission from passenger car and motorcycle engines. Journal of Chromatographic Science 33, 34}48. McLaren, R., Singleton, D.L., Lai, J.Y.K., Khouw, B., Singer, E., Wu, Z., Niki, H., 1996. Analysis of motor vehicle sources and their contribution to ambient hydrocarbon distributions at urban sites in Toronto during the Southern oxidants study. Atmospheric Environment 30 (12), 2219}2232. Simpson, D., et al., 1999. Inventorying emissions from nature in Europe. Journal of Geophysical Research 104 (7), 8113}8152. Solberg, S., Dye, C., Schmidbauer, N., Herzog, A., Gehrig, R., 1996. Carbonyls and non-methane hydrocarbons at rural sites from the Mediterranean to the Arctic. Journal of Atmospheric Chemistry 25, 33}66. Steinbrecher, R., 1998. personal communication. Ye, Y., Galbally, I.E., Weeks, I.A., 1997. Emission of 1,3-Butadiene from petrol-driven motor vehicles. Atmospheric Environment 31 (8), 1157}1165.