application and development of surface layer flux ... - E-thesis

FINNISH METEOROLOGICAL INSTITUTE CONTRIBUTIONS No. 31

APPLICATION AND DEVELOPMENT OF SURFACE LAYER FLUX TECHNIQUES FOR MEASUREMENTS OF VOLATILE ORGANIC COMPOUND EMISSIONS FROM VEGETATION Janne Rinne

ACADEMIC DISSERTATION in meteorology University of Helsinki, Faculty of Science, Department of Physical Sciences To be presented, with the permission of the Faculty of Science of the University of Helsinki, for public criticism in Auditorium Porthania IV (Yliopistonkatu 3) on September 1, 2001, at 10 a.m.

Finnish Meteorological Institute Helsinki, 2001

2

This version: ISBN 952-10-0101-1

Printed version: ISBN 951-697-545-3 ISSN 0782-6117

Helsinki 2001

3

Abstract

Volatile organic compound (VOC) emissions from vegetation have been estimated to be several times higher than anthropogenic emissions. These estimates, however, have large uncertainties due to uncertainties in the emission parameters, emission algorithms and vegetation information used in the model calculations. In this work surface layer flux measurement methods were developed and deployed to measure VOC emissions from vegetation and to obtain emission parameters and test algorithms at canopy scale. Chemical sampling and analysis proved to be one of the most important sources of uncertainty for the surface layer flux techniques for VOCs. The parameterizations used for the turbulent exchange coefficients in the gradient technique with Monin-Obukhov stability functions, and the roughness sub-layer can add systematic uncertainty to the fluxes presented. Two eddy flux systems based on the disjunct eddy sampling (DES) approach were developed for measurement of VOC fluxes: a disjunct true eddy accumulation and a disjunct eddy covariance system. In the DES, air samples are taken within a very short time (0.2 s) but with a longer interval between them. This reduces the number of samples used to obtain the flux but gives more time to process the samples. This makes eddy covariance measurements possible using relatively slow sensors, such as the proton-transfer-reaction mass spectrometer used in this work. The DES approach bypasses some technical difficulties of the true eddy accumulation method, in which the sampling rate is proportional to the vertical wind velocity. The flux measurements above boreal Scots pine and Siberian spruce/mountain birch forests showed that the monoterpene emissions from this type of forest are probably slightly overestimated in emission models. Isoprene fluxes from these forests were below detection limit. The monoterpene fluxes were temperaturedependent as expected, but this temperature dependence was somewhat higher than obtained using enclosure techniques. The measured VOC fluxes above primary tropical rainforest in Brazil showed high fluxes of isoprene and lower fluxes of monoterpenes. The monoterpene fluxes above rainforest showed a light and temperature dependent behavior similar to the one observed usually for isoprene emissions. High methanol fluxes were measured above an undisturbed alfalfa field.

4

5

KIITOKSET

Kiittää kaikkia väitöskirjaan liittyneiden artikkeleiden tekemiseen osallistuneita, sillä tieteenteko on usein ryhmätyötä. Kiitän myös muita työtovereitani Ilmatieteen laitoksen Troposfäärin otsonitutkimusryhmässä ja National Center for Atmospheric Research:in Biosphere-Atmosphere Interactions projektissa. Tuomas Laurila IL:lta ja Alex Guenther NCAR:stä ansaitsevat erityiskiitokset hedelmällisen tutkimusilmapiirin luomisesta ja ylläpitämisestä. Kiitän myös ohjaajaani Sylvain Joffrea Ilmatieteen laitokselta työhön liittyvistä neuvoista ja tuesta. Kiitän Maj ja Tor Nesslingin säätiötä, Suomen Akatemiaa, National Center for Atmospheric Research:ia ja Ilmatieteen laitosta työni rahoituksesta. Lopuksi haluaisin kiittää ystäviäni ja sukulaisiani, jotka estivät minua käyttämästä kaikkea aikaani työntekoon.

ACKNOWLEDGEMENTS

First I want to thank all my co-authors, for science is nowadays usually collective work. I thank all my colleagues at Tropospheric Ozone research group of the Finnish Meteorological Institute in Helsinki, and at the Biosphere-Atmosphere Interactions project of the National Center for Atmospheric Research in Boulder, Colorado. Tuomas Laurila of FMI and Alex Guenther of NCAR deserve special thanks for creating and keeping up creative and supportive work environment. I also thank my supervisor Sylvain Joffre of FMI for his support and advice during this work. I am grateful for Maj and Tor Nessling Foundation, Academy of Finland, National Center for Atmospheric Research, and Finnish Meteorological Institute for their financial support. Finally I thank my friends and family for keeping me from working all the time.

Janne Rinne Boulder, Colorado June 15, 2001

6

PAPERS INCLUDED IN THE THESIS

I: Rinne, J., Tuovinen, J.-P., Laurila, T., Hakola, H., Aurela, M. & Hypén, H., 2000. Measurements of hydrocarbon fluxes by a gradient method above a northern boreal forest. Agricultural and Forest Meteorology, 102, 25-37.

II: Rinne, J., Hakola, H., Laurila, T. & Rannik, Ü., 2000. Canopy scale monoterpene emissions of Pinus sylvestris dominated forests. Atmospheric Environment, 34, 10991107.

III: Rinne, H. J. I., Delany, A. C., Greenberg, J. P. & Guenther, A. B., 2000. A True eddy accumulation system for trace gas fluxes using disjunct eddy sampling method. Journal of Geophysical Research, 105, 24791-24798.

IV: Rinne, H. J. I., Guenther, A. B., Greenberg, J. P. & Harley, P. C., 2001: Isoprene and monoterpene fluxes measured above Amazonian rainforest and their dependence on light and temperature. Submitted for publication in Atmospheric Environment.

V: Rinne, H. J. I., Guenther, A. B., Warneke, C., de Gouw, J. A. & Luxembourg, S. L., 2001: Disjunct eddy covariance technique for trace gas flux measurements. Geophysical Research Letters, in press.

7

CONTENTS 1 INTRODUCTION ....................................................................................................................................... 9 2 BIOGENIC VOCs IN PLANTS AND THE ATMOSPHERE .................................................................. 12 3 METHODS FOR ESTIMATING BIOGENIC VOC EMISSIONS........................................................... 17 3.1 Enclosure techniques .......................................................................................................................... 18 3.2 Surface layer flux measurement techniques........................................................................................ 19 3.2.1 Eddy covariance method.............................................................................................................. 22 3.2.2 Surface layer gradient methods.................................................................................................... 24 3.2.3 Eddy accumulation methods ........................................................................................................ 27 3.2.4 Disjunct eddy sampling................................................................................................................ 29 3.3 Mixed layer techniques ....................................................................................................................... 30 3.4 Sampling and chemical analysis of VOCs .......................................................................................... 31 3.5 Emission modeling.............................................................................................................................. 35 4 RESULTS .................................................................................................................................................. 37 4.1 Gradient technique .............................................................................................................................. 38 4.2 Disjunct eddy sampling....................................................................................................................... 39 4.3 Canopy-scale VOC emissions............................................................................................................. 40 4.4 Implications for future research .......................................................................................................... 47 5 SUMMARY AND CONCLUSIONS ........................................................................................................ 49 REFERENCES ............................................................................................................................................. 51 APPENDIX A............................................................................................................................................... 67 APPENDIX B ............................................................................................................................................... 69 PAPERS I-V

8

9

1 INTRODUCTION

Biological processes play an important role in controlling the chemical composition of the atmosphere. The abundance of oxygen, and subsequently the existence of the ozone layer, is due to photosynthesis. Many greenhouse gases, such as carbon dioxide and methane, are at least partly controlled by the biosphere. Furthermore, vegetation also partly controls the hydrological cycle and emits chemically reactive trace gases into the atmosphere. Boreal forests and tropical rainforests are the major forested biomes (11 % and 5 % of land surface area respectively) affecting the composition of atmosphere (Figure 1).

More than forty years ago Dr F. W. Went published his ideas on the role of vegetation on the origin of blue hazes in remote vegetated areas. He reasoned that these hazes are formed by condensation of organic compounds originating from vegetation (Went 1955; 1960). The biogenic volatile organic compounds (BVOCs) were studied intensively during seventies, but almost forgotten in early eighties. In the late eighties and early nineties the role and importance of biogenic VOC emissions into the atmosphere became again high on the scientific agenda for several issues including tropospheric ozone formation and climate change. In the past decade biogenic volatile organic compounds have been under intense research. This research has been reviewed by Fuentes et al. (2000). Early research on BVOCs is reviewed by Rasmussen (1981) and by Lerdau et al. (1997).

10

Figure 1. Boreal forests (light gray) and tropical rain forests (dark gray) as derived from AVHRR NDVI data (Documentation available at http://edcdaac.usgs.gov/glcc/globdoc1_2.html).

According to emission estimates, the global biogenic VOC emissions are many times higher than the anthropogenic emissions (Müller, 1992; Guenther et al., 1995). Also on the regional scale there are areas, even in Europe, where biogenic emissions are higher than anthropogenic ones, for example Finland (Simpson et al., 1999; Lindfors and Laurila 2000). However, BVOC emission inventories have large uncertainties since the emission potentials of many plant species are not known, and the up-scaling procedure from leaf level to larger scales is an important source of uncertainty.

Emission models require information on the emission potentials of vegetation. There exists several methods for obtaining this information on different spatial and temporal scales. Enclosure, (or cuvette, or chamber) methods are used at the leaf and branch scale. Micrometeorological surface layer flux measurement methods give emissions at the canopy scale. The boundary layer mass balance and gradient methods provide landscape scale emissions. Each of these methods can be used to obtain the emission parameters for

11

emission models or to validate model results based on emission factors obtained by another method.

The papers included in this thesis deal with surface layer flux measurements of BVOCs. The main goals of the work were to obtain measurement data of canopy-scale BVOC emissions of different vegetation types, to identify and quantify the sources of uncertainty associated with these measurement techniques, and to further develop techniques used to measure BVOC emissions in the canopy scale. The work consists of applying existing methods, analyzing the uncertainties associated with these measurements, developing new measurement techniques, and analyzing the data obtained by these techniques.

Papers I and II deal with BVOC fluxes measured by a gradient technique with turbulent exchange coefficients derived using the Monin-Obukhov similarity theory and the Businger-Dyer formulae. In paper I sources of uncertainty in gradient measurements are described in detail. The results of the uncertainty analysis are used in paper II. In paper II results from monoterpene flux measurements over Scots pine (Pinus sylvestris) forests are presented and compared to model estimates.

Papers III and V concentrate on method and instrument development using a new approach, disjunct eddy sampling. This approach is incorporated to the true eddy accumulation (paper III) and the eddy covariance (paper V) method. In paper IV VOC fluxes measured by the disjunct eddy accumulation method above a neo-tropical rainforest are presented.

12

The work presented in papers I and II was conducted as a part of BIPHOREP research project (Laurila and Lindfors, 1999) and that presented in paper IV as a part of the LBAEcology project (http://lba-ecology.gsfc.nasa.gov/lbaeco/).

The author of this thesis bore the main responsibility of the measurements and data analysis, with the exception of chemical analysis of the air samples (papers I-IV), which was conducted by H. Hakola and J. Greenberg, and operation of the fast isoprene sensor and analysis that data (paper IV), which was conducted by A. Guenther. The PTR-MS used in paper V was operated by C. Warneke and S. Luxembourg. Author also bore the main responsibility of the instrument development described in papers III and V.

2 BIOGENIC VOCs IN PLANTS AND THE ATMOSPHERE

In many remote vegetated areas, concentrations of volatile organic compounds (VOC), scaled by the reactivity against ozone and hydroxyl radical, are dominated by biogenic compounds (see Figure 2) (e.g., Laurila and Hakola, 1996; Hakola et al., 2000). Above tropical rainforests the BVOC concentrations are most often dominated by isoprene (C5H8, 2-methyl-1,3-butadiene) (Zimmerman et al., 1988; Helmig et al., 1998; Greenberg et al., 1999a; Kesselmeier et al., 2000). On the other hand, in the European boreal zone

13

Monoterpenes Isoprene Light alkanes (C2-C5) Light alkenes (C2-C5) benzene & acetylene

Propy-Eq. [ppbC]

16

12

8

4

0 May-97

Jun-97

Jul-97

Aug-97

Sep-97

Oct-97

Propy-Eq. [ppbC]

16

12

8

4

0 Apr-98

May-98

Jun-98

Jul-98

Aug-98

Sep-98

Oct-98

Figure 2: OH-reactivity scaled monthly mean VOC concentrations at Pötsönvaara, Ilomantsi, Finland according to Hakola et al. (2000).

the monoterpene (C10H16) concentrations commonly exceed the isoprene concentrations (e.g. Hakola et al., 2000). α-Pinene ((1S)-2,6,6-Trimethylbicyclo[3.1.1]hept-2-ene) is

14

often the dominant monoterpene, other common monoterpenes being β-pinene, ∆3-carene and limonene.

In the atmosphere the degradation of un-saturated hydrocarbons, such as isoprene and monoterpenes, starts usually with addition of OH, O3 or NO3 to a C=C double bond. The atmospheric chemistry of these hydrocarbons is complex and beyond the scope of this thesis. Depending on atmospheric conditions the degradation of BVOCs can lead to various end-product combinations. The common products of isoprene degradation include formaldehyde, methacrolein and methyl-vinyl-ketone (Seinfeld and Pandis, 1998). Products of α-pinene, which is often the most common monoterpene in the air, include pinonaldehyde, formaldehyde and acetone (Hakola et al., 1994; Nozière et al., 1999).

In the presence of NOx, hydrocarbons tend to shift the NO to NO2 balance towards NO2, thus increasing ozone production, if sufficient NOx is available (Seinfeld and Pandis, 1998; Atkinson, 2000). Much of the interest in BVOCs has arisen from their capacity to affect the ozone chemistry in the boundary layer (Chameides et al., 1992; Fehsenfeld et al., 1992). Early ozone reduction measures were often unsuccessful because they did not take BVOCs into account (NRC, 1991).

Recently, the capacity of BVOCs to take part in aerosol formation and growth has increased the interest in BVOCs (Kavouras et al., 1998; Kulmala et al., 1998, 2000; Griffin et al., 1999; Andersson-Sköld and Simpson, 2001). Degradation products of

15

monoterpenes have been observed in laboratory studies to contribute to aerosol formation and growth. However, initial monoterpene concentrations in these studies have been orders of magnitude higher than those observed in nature (Hoffmann et al., 1997; Nozière et al., 1999).

The tropical atmosphere, due to the intense radiation and high humidity, is the major region for oxidation of such powerful greenhouse gases as methane (Crutzen and Zimmerman, 1991). As reactive organic compounds affect the oxidation capacity of the tropical atmosphere, they can have an indirect effect on the global greenhouse effect via methane oxidation. Biogenic hydrocarbons are also important for the budget of carbon monoxide in the troposphere (Granier et al., 2000). Compared to global carbon dioxide fluxes, the global BVOC emission (1150 TgC/yr, Guenther et al., 1995) is minuscule, but still the same order of magnitude as the rate of increase of carbon dioxide in the atmosphere (3300 TgC/yr, IPCC, 1996).

Plants produce volatile organic compounds for various purposes, e.g. ethene acts as growth hormone; while many monoterpenes serve as defense against herbivores (Tingey et al., 1991; Langenheim, 1994, and references therein). The purpose of isoprene, which is the most studied atmospheric BVOC, is not known yet. It has been suggested that plants synthesize isoprene for thermal protection (Sharkey and Singsaas, 1995), or that isoprene scavenges gaseous nutrients, such as nitrous oxides, from the air to the forest soil (Klinger et al., 1998). Recently it has been suggested that isoprene production participates in the regulation of chloroplast metabolism (Logan et al., 2000).

16

The VOC emission mechanisms and the dependence of emissions on environmental parameters varies from compound to compound and also between plant species. Isoprene is emitted directly from synthesis and its emission depends on both light and temperature (Guenther et al., 1991). Many plants emit monoterpenes from storage pools and their emission is temperature dependent (Tingey et al., 1991; Guenther et al., 1991). However, it has been shown that some Mediterranean and tropical plants emit monoterpenes in light and temperature dependent fashion directly from synthesis (Staudt and Seufert, 1995; Loreto et al., 1996; Kuhn et al., 2001). Vegetation also emits several other VOCs, such as methanol and hexenal. The emissions of certain compounds can be greatly influenced by, e.g., drought stress, flowering and insect damage (Hansen and Seufert, 1999; Litvak et al., 1999).

Isoprene is emitted by many, mainly deciduous broadleaved, trees, such as oaks (Quercus spp.), poplars and aspens (Populus spp.) and willows (Salix spp.). Many tropical tree species also emit isoprene. Many evergreen coniferous trees such as pines (Pinus spp.) and spruces (Picea spp.) are monoterpene emitters. As the biogenic emissions of hydrocarbons are highly dependent on plant species, a change in land-use can have a significant effect on regional BVOC emissions (Guenther et al., 1999; Schaab et al., 2000).

Currently the main scientific issues driving the study of biogenic VOC emissions include the role of BVOCs in ozone production, in the oxidant balance of the troposphere, and in

17

aerosol formation. The characterization of emissions from tropical vegetation is of great importance for our knowledge on the global oxidation capacity of the troposphere, as this area is important for the oxidation of such powerful greenhouse gas as methane, and so little is still known about BVOC emissions in the tropics. The effect of land-use change on the BVOC emissions can have a great impact on ozone formation in many areas in the industrialized world where reforestation commonly occurs. Deforestation and forest degradation in the tropics can have an effect on the oxidation capacity in these areas. Very little is still known about the oxygenated VOC emissions from vegetation, which can be important.

3 METHODS FOR ESTIMATING BIOGENIC VOC EMISSIONS

There exist a variety of methods for estimating emissions of VOCs from the vegetation at various spatial and temporal scales. The measurement methods include enclosure, surface layer flux and mixed layer techniques described below and reviewed by Dabberdt et al. (1993). All these methods rely on our ability to measure the concentrations of trace gases accurately enough. Results of emission measurements are often used as an input data for emission models or to validate model results. The theory behind the surface layer flux techniques and the mixed layer techniques is described by, e.g., Stull (1988), Garratt (1994), and Kaimal and Finnigan (1994).

18

3.1 Enclosure techniques

Enclosure, or cuvette, techniques are used to measure gas exchange between plants and the atmosphere at the leaf or branch scale. In these techniques, a leaf or a branch, or occasionally a whole tree, is enclosed into a cuvette. With dynamic cuvettes, air is pumped through the cuvette and the concentration difference between incoming and outgoing air is used to calculate the emission from the plant tissue inside. With static cuvettes air is not pumped through and the rate of change of the concentration is used to determine the emission or assimilation rates.

Dynamic cuvettes are widely used to measure VOC emissions from plants. Emissions of various plant species have been measured with cuvette techniques by, e.g., Rasmussen (1972), Tingey et al. (1979), Evans et al. (1985), Isidorov et al. (1985), Juuti et al. (1990); Guenther et al. (1991, 1996), Janson (1993), Hakola et al. (1998) Janson et al. (1999), Geron et al. (2000), Boissard et al. (2001) and Pétron et al. (2001). Enclosure methods are an important tool for screening the emissions from different plant species and for studying the dependencies of VOC emissions on environmental parameters.

One of the main advantages of the enclosure methods is their simplicity. However, since these methods give the emission parameters at the leaf to tree scale, the subsequent upscaling can introduce large uncertainties into the results. Moreover, the enclosure itself affects the environment around the enclosed part of the plant. This can have an unknown

19

effect on the measured gas exchange. Too rough handling of the plant can also lead to artificially high emissions (Juuti et al., 1990).

3.2 Surface layer flux measurement techniques

Micrometeorological surface layer flux (SLF) methods can be used to measure VOC emissions from vegetation at the canopy scale. The surface layer of the atmosphere is the lowest part of the atmospheric boundary layer, in which the effect of the coriolis force on the flow is negligible. In these methods, measurements are conducted above a vegetation canopy, and the measured emission fluxes represent average emissions from an area upwind from the measurement point, called the flux footprint, or source area. Depending on the height of the vegetation and the structure of the canopy, the measurement height can vary from about a meter to few tens of meters above the canopy.

The SLF measurement methods impose several requirements to the measurement site, the measured compounds and on atmospheric conditions. The surface around the measurement point must be reasonably flat and horizontally homogenous both in roughness and source strength. To obtain reliable flux values, surface layer methods require strong enough turbulence and quasi-stationary conditions, i.e., that the turbulence statistics approach stable value as averaging time is increased (see, e.g., Kaimal and Finnigan, 1994). Also the chemical lifetime of the compound measured must be much

20

Figure 3. A schematic figure of two-dimensional footprint function with measurement point at [0,0] (black dot) and the mean wind direction indicated by the arrow. The footprint function is calculated according to Horst (1999) with following parameters z=2.6 m, L=-10 m, u*=0.5 m/s.

longer than the timescale of turbulent mixing, i.e., the Damköhler number (see chapter 3.2.2 Surface layer gradient methods) must

be small.

The area upwind from the measurement point, which affects the measured flux, is called the footprint. This is not a well-defined area but rather is a continuous impact function, which describes how strongly an emission from a specific surface unit affects the flux measured downwind (Figure 3). The location and extent of the footprint strongly depends on atmospheric stability. Various estimates of the footprint area, based on analytical solutions of diffusion equations or numerical simulations, are described by, e.g., Schuepp

21

et al. (1990), Schmid (1994), Horst (1999) and Rannik et al. (2000). Since a basic requirement for SLF measurements is horizontal homogeneity of the surface, footprint analysis is an important part of the assessment of the data quality. For example, according to Schuepp et al. (1990), the 75 % cumulative footprint in hydrostatically neutral conditions extends to 1000 meters from a measurement point at a height of 22 meters, and to 300 meters from a measurement point at 9 meters height. Under hydrostatically unstable conditions, typical for daytime, the cumulative footprints are shorter.

The sources of uncertainty and accuracy needed for concentration measurements required for SLF measurements have been studied by, e.g., Wesely and Hart (1985), Businger (1986) and Businger and Delany (1990). Compared to the enclosure measurements the SLF methods require much more accurate concentration measurements. Moreover, even without taking the instrumental uncertainties into account, the natural random variation in the turbulence field introduces typically an uncertainty of 10-20% into flux value averaged over half an hour. On the other hand, SLF methods measure spatially averaged, canopy scale, fluxes with minimum disturbance on the vegetation and its environment.

Since SLF measurements must be conducted above the canopy, they require developed infrastructure, especially at forested sites where towers several tens of meters of height are needed. These methods often require also enough electricity and relatively easy access to the measurement site.

22

Intercomparison studies of various methods for BVOC flux measurements are still scarce. Darmais et al. (2000) compared monoterpene fluxes measured by a relaxed eddy accumulation and gradient method and found a reasonably good agreement between those two methods. Gallagher et al. (2000) intercompared CO2 fluxes measured by an REA system for VOCs with an eddy covariance system and found a good agreement.

3.2.1 Eddy covariance method The most direct method for surface layer flux measurements is the eddy covariance method (EC). It requires high-frequency simultaneous measurements of concentration of a compound (c) and vertical wind velocity (w). The covariance between the eddy components of these variables ( c' = c − c , w' = w − w ) gives the flux (Fc): t

1 2 Fc = w' c' = w' (t )c' (t )dt , t 2 − t1 ∫t1

(1)

where the overbar denotes time-averaging. Since a substantial share of the fluxes in the atmospheric surface layer is transported by relatively small eddies, the measurements of w and c must be made with fast-response sensors. Typical sampling frequencies used in EC measurements are 5-10 Hz. The three dimensional wind is commonly measured by an acoustic anemometer wherein each component of the wind (u, v, w) is obtained from the differences in time it takes for an acoustic signal to travel the same path in opposite directions. The virtual temperature is obtained simultaneously from the speed of sound.

The EC method is commonly used for measurements of CO2, H2O and O3 fluxes. Some instruments for measuring fluxes of isoprene (Guenther and Hills, 1998) and aerosols

23

(Buzorius et al., 1998) by EC method have also been developed. However, for many atmospheric constituents there exist no sensors fast enough for traditional EC measurements, although recently the proton transfer reaction mass spectrometry has been applied for EC measurement of several hydrocarbons (Karl et al., 2001).

Even though the eddy covariance method is the most direct SLF method, some corrections are necessary to the fluxes as defined by Equation (1). These include correction taking into account density fluctuations caused by temperature and humidity variations into account, non-sufficient frequency response of sensors, loss of low frequencies due to finite sampling period, as well as the displacement of sensors (Webb et al., 1980; Moore, 1986; Lenschow et al., 1994; Kristensen et al., 1997; Massman, 2000). Also slight deviations from ideal conditions, such as non-stationarity and non-horizontal mean wind can be corrected for (McMillen, 1988). Methods for correcting the advective effects have recently been presented (Paw U et al., 2000).

Advantages of the eddy covariance method include also the possibility to post-process the raw turbulence data, which makes experimentation with different data processing methods feasible. Eddy covariance techniques, however, usually require quite developed infrastructure for the measurement site, and are often used at established long-term research sites.

24

3.2.2 Surface layer gradient methods Surface layer gradient techniques are the traditional methods for measuring fluxes of trace gases, for which no fast sensor exists. They are based on the assumption that the flux can be parametrised, in analogy with molecular diffusion, through the vertical gradient of the compound in question, and a turbulent exchange coefficient Kc, i.e.: Fc = − K c

∂c . ∂z

(2)

This equation is a first order local turbulence closure and it assumes that the major part of the turbulent transport is dominated by eddies smaller than the distance over which the gradient is measured. This requirement is not met with elevated measurement heights in a strongly convective boundary layer. Since the turbulent exchange coefficient is a property of the flow and not of the compound transported, a similarity between exchange coefficients for passive scalar properties can be assumed. However, close to vertically distributed source or sink the similarity between transportation of scalars with different vertical source distributions can break down.

The turbulent exchange coefficient itself can be estimated in several ways. The most direct method is to invert equation (2), and use another scalar, for example temperature or H2O, as a tracer. This method, called the modified Bowen-ration method, requires very accurate gradient measurements of the tracer and can induce a large error into calculated flux values. Moreover, as the gradient goes to zero with the flux, the turbulent exchange coefficient is not well defined under these conditions by the modified Bowen-ratio method.

25

Another method to estimate Kc is to use the Monin-Obukhov similarity theory (see, e.g., Stull, 1988; Garratt, 1994; Kaimal and Finnigan, 1994) and the Businger-Dyer formulae (Businger et al., 1971; Dyer, 1974), or equivalent ones, for dimensionless stability functions, as is done in papers I-II. This can be expressed as: Kc = Kh =

κu ∗ z , φ h (ζ )

(3)

where Kh is the turbulent exchange coefficient for heat, κ is the von Kármán constant, u∗ is the friction velocity and φh is the dimensionless stability correction function for heat transfer and depends on the Monin-Obukhov stability parameter ζ. Equation (3) assumes a similarity between heat and mass transport. The full derivation of the equation used for flux calculations in papers I-II (Eq. (2) in paper I, Eq. (1) in paper II) is described in Appendix A., based on Garratt (1994) and Fuentes et al. (1996). This similarity theory approach does not require demanding measurements of a tracer gradient. It, however, can introduce systematic errors into flux measurements.

Use of the gradient method as described above requires the timescale of turbulent mixing

τ∗ to be much shorter than the timescale of the chemical reactions involved τch, i.e. the Damköhler number Da=τ∗/τch must be small. Formulations for gradient-flux relations in the cases where the Damköhler number is not small have also been developed (VilàGuerau de Arellano et al., 1995).

The conventional flux-gradient relationships tend to break down near a very rough surface, such as a forest canopy (Garratt, 1980). Equation (3) alone does not take this so-

26

called roughness sub-layer (RSL) effect into account. This can lead to serious underestimation of fluxes, since the gradient within the RSL is smaller than the one predicted by flux-gradient relationships. The underestimation of the fluxes due to the RSL is estimated to be between 0 and 70 % depending on the ratio of the measurement height to the canopy height (Simpson et al. 1998, and references therein). A RSL correction can be applied to compensate for this effect. To obtain a numerical value for this correction, one can study relationships between measured carbon dioxide, water vapor or temperature gradients, and fluxes measured by the eddy covariance method. In the cases where there are no gradient measurements accurate enough for this, relations between RSL correction and characteristics of the surface roughness can be used (Garratt, 1980; Cellier and Brunet, 1992). However, there exists no consensus on the formulation and magnitude of these corrections.

The most convenient and reliable way of using the gradient method for trace gas fluxes is to use the modified Bowen-ratio and the Monin-Obukhov similarity approach together. The turbulent exchange coefficients or dimensionless gradients can be obtained using both techniques, thus making it possible to define RSL correction functions for the site. For trace gas flux calculations, exchange coefficients calculated using the MoninObukhov similarity approach are then used, but corrected for RSL. This method allows trace gas flux measurements even when the gradients of other scalars are near zero.

27

Because gradient measurements involve concentration measurements conducted at different heights, they impose more strict requirements for horizontal homogeneity of source distribution, since different measurement heights have different footprints.

Gradient techniques have been used for BVOC flux measurements by, e.g., Lamb et al. (1985), Fuentes et al. (1996), Goldstein et al. (1996), Guenther et al. (1996), Cao et al. (1997), Schween et al. (1997) and Darmais et al. (2000).

3.2.3 Eddy accumulation methods Eddy accumulation techniques are more direct methods for flux measurements of trace gases than the gradient method. They imply that air samples are accumulated into updraft and down-draft reservoirs using data from an acoustic anemometer to control the sampling. In the original true eddy accumulation technique, the sampled volume is proportional to the magnitude of the vertical wind velocity (Desjardins, 1977) and Fc = w + c ↑ + w − c ↓ ,

(4)

where w + is the time average of updrafts and w − that of downdrafts, as defined in Appendix B. The quantity c↑ is the concentration of a trace gas in an updraft sample, and c↓ that in a downdraft sample. Equation (4) can be derived from equation (1) as is described in Appendix B. However, difficulties in controlling the sampling flow accurately and fast enough have hindered the use of this method.

Instead, the relaxed eddy accumulation method (REA) has gained popularity for BVOC flux measurements. In this method, introduced by Businger and Oncley (1990) the

28

sampling flow is kept constant. Additionally, an optional dead-band around zero vertical wind velocity was introduced to increase the concentration difference between up- and downdraft samples. Thus, the flux is defined as

Fc = σ w b(c ↑ − c ↓ ) ,

(5)

where σw is the standard deviation of the vertical wind velocity and b a theoretical or empirical constant. The latter is needed since information is lost due to constant sampling-rate. Theoretically b depends on the joint probability distribution of w and c, and has a value of 0.627 if this is Gaussian (Baker et al., 1992). Empirical values for b are usually around 0.6 (e.g., Businger and Oncley, 1990; Baker et al., 1992). The value of b can also be derived from high frequency data of another scalar, e.g. temperature, by inverting equation (5). Then a similarity between scalar transport is again assumed.

The REA technique has been a method of choice for BVOC flux measurements for the past few years. Its advantages include the fact that the samples can be transported even to another continent for analysis, thus not requiring a laboratory nearby the measurement site. The same holds true also for gradient method. It is also possible to build REA systems using very little power, making measurements independent of line power sources. As the REA technique requires on-line decisions on the up- or downdraft sampling, the non-horizontal mean wind cannot be corrected during post-processing of the data. This makes it less applicable for measurements at non-ideal sites. The REA technique has been used for BVOC flux measurements by e.g. Guenther et al. (1996), Valentini et al. (1997), Baker et al. (1999), Ciccioli et al. (1999), Christensen et al. (2000) and Darmais et al. (2000).

29

3.2.4 Disjunct eddy sampling All measurement techniques mentioned above usually sample in a continuous manner. Since fluxes can be reliably calculated also from a subset of a continuous time series, a disjunct sampling (DES) method can be used (Haugen, 1978; Kaimal and Gaynor, 1983; Dabberdt et al., 1993; Lenschow et al., 1994). In this approach, short separate samples are taken from the continuous time series, and ensemble averaging them yields the flux. In the case of the eddy covariance method the flux would be an ensemble-averaged covariance: Fc = w' c' =

1 N

N

∑ w' (t )c' (t ) , i =1

i

i

(6)

instead of a time averaged covariance (Eq. 1). Techniques based on the disjunct eddy sampling approach are described in papers III and V, and applied in paper IV. The DES gives more time to process each sample than continuous sampling, but it also reduces the number of samples considerably. As the time-series of any atmospheric property in the surface layer are autocorrelated, the disjunct sampling does not increase the uncertainty as much as would be the case with random time series. According to Lenschow et al. (1994), the additional uncertainty introduced to a flux value due to disjunct sampling is less than 8 % if the time-distance between samples is less than the appropriate integral time-scale. The major advantage of DES is that it makes direct eddy covariance flux measurements possible using analyzers with response times between one and 30 seconds, and maybe even longer. Even when there exists a fast analyzer for a compound, a slower instrument can be smaller, more stable, easier to operate, and less expensive.

30

3.3 Mixed layer techniques

Mixed boundary layer methods are used to estimate BVOC emissions at a scale of several square kilometers. They require concentration measurements at several heights within the well-mixed boundary layer and are only applicable under convective conditions. Tethered balloons are commonly used as platforms for the measurements, but also aircraft can be used.

In the mixed layer gradient method the vertical profile of a conserved compound is assumed to be the result of a superposition of bottom-up and top-down fluxes, Fc0 and Fch: F F ∂c = − c 0 g b ( z h) − ch g t ( z h) , ∂z w∗ h w∗ h

(7)

where w∗ is the convective velocity scale, h the boundary layer depth and gb and gt are dimensionless gradients for bottom-up and top-down diffusion, respectively (Wyngaard and Brost, 1984). The dimensionless gradient terms can be estimated using large eddy simulations.

The boundary layer budget equation for a compound can be written as: F − Fch ∂C = U ⋅∇C + c 0 +Q, ∂t h

(8)

where C is the concentration of the compound integrated through the whole boundary layer. The first term on the right hand side of equation (8) is the horizontal advection term, the second the vertical flux divergence within the boundary layer, and the third term

31

Q describes sources and sinks in the boundary layer (see, e.g., Garratt, 1994). The emission of a reactive trace gas, Fc0, can be derived by Equation (8) by measuring the time change of its concentration integrated throughout the mixed boundary layer, usually neglecting advection, and modeling entrainment flux Fch and net chemical production Q.

Mixed layer methods have been used for BVOC emission estimates by e.g. Zimmerman et al. (1988), Guenther et al. (1996), Helmig et al. (1998) and Greenberg et al. (1999a, 1999b). Mixed layer methods can give the VOC emissions on a relatively large scale thus being important tools for validating the up-scaling procedures of the emissions. These methods, however, rely on modeling the complex effect of chemistry and entrainment flux on the profiles and the errors involved can lead to large uncertainties in the estimated emissions.

3.4 Sampling and chemical analysis of VOCs

The reliability of the BVOC flux measurements strongly depends on our ability to accurately sample and analyze atmospheric VOC concentrations. Sampling and analysis methods have been reviewed by Cao and Hewitt (1999). The methods used in this work are discussed in this section.

In most methods for VOC flux measurements, compounds are first sampled into a storage reservoir to be analyzed later. Sampling methods include whole air sampling and adsorbent techniques. Whole air sampling implies sampling air into a bag or a canister.

32

Rigid canisters are usually used if samples are not analyzed on site since they are easier to store and are more rugged for transportation. The inside of the storage volume has to be non-reactive with target compounds and not emitting compounds that might lead to sample contamination. Passivated stainless steel canisters are generally used for VOCs up to C10 compounds.

Adsorbent techniques are a convenient way for VOC sampling since adsorbent cartridges are usually small in size and, if not made of glass, rugged in use. In these techniques, air is drawn through a tube filled with one or more solid adsorbents. As the air moves through, hydrocarbons stick onto the surface of the adsorbent. As different materials adsorb different compounds, the right adsorbent for target compounds has to be selected. Two or more adsorbents can also be used together so that air flows first through the weaker adsorbent that adsorbs the larger molecules, and then through the stronger one, adsorbing smaller molecules. All adsorbents have breakthrough volume after which the adsorbent is not able to totally adsorb the compounds. Also chemical conversion of compounds can occur on the adsorbent.

There exist a variety of adsorbent materials, each with their own advantages and disadvantages. Adsorbents used in the work described in this thesis were Tenax TA, Carbotrap and Carbosieve. Tenax TA is an organic polymeric resin which can be used to collect monoterpenes and other hydrocarbons larger than C6. Carbotrap and Carbosieve are carbon-based adsorbents, which can be used together to collect C3-C12 hydrocarbons.

33

The compounds are desorbed for analysis from adsorbent by heating it up to around 300ºC.

As sampled air flowing through the adsorbent contains also ozone, which can react with hydrocarbons, ozone removal can be necessary at high ambient ozone levels. No ozone scrubbing was used in the studies described in papers I, III and IV. The highest ozone concentration in Northern Finland during the measurements described in paper I were around 40 ppbv. In Niwot Ridge (Paper III) the ozone concentrations were around 50 ppbv. In Amazon region (paper IV) ozone concentrations are usually below 20 ppbv, except during the biomass burning season (Cordova Leal et al., 2000; Vanni Gatti et al., 2000). In the work described in paper II, MnO2 coated Cu mesh was used to remove ozone.

The samples taken following the above mentioned ways are usually analyzed using gas chromatography (GC) methods. For GC methods the sample is usually first preconcentrated using a cryotrap and then flushed into a column wherein compounds are separated on the basis of their different partition coefficients.

After separation, compounds are measured by a suitable method, such as flame ionization detection (FID) or mass spectrometry (MS). In FID, compounds coming from the column are ionized by a hot flame. These ions are then accelerated towards a sensor and the resulting current is recorded. Thus, the compounds are identified only by their retention times in the column. In mass spectrometry compounds are also ionized, often by electron

34

impact. The produced ions are directed toward an exit slit into the mass analyzer, wherein different masses are directed to the detector using a varying electrical field. In the scan mode all of the masses are observed. Better accuracy and lower detection limit can be obtained by using a selective ion monitoring (SIM) mode. With the SIM mode, only preselected ions are observed thus giving longer integration time but little possibilities for compound identification.

There exist some instruments for fast measurement of volatile organic compounds. Guenther

and

Hills

(1998) developed

a

fast

isoprene

analyzer

based

on

chemiluminescence of degradation products of isoprene and ozone with response time of 0.5 s. However, there are interferences with some other compounds, such as propene. This instrument is thus best suited for flux measurements in environments where isoprene is the dominant VOC.

Proton-transfer-reaction mass spectroscopy (PTR-MS) is a new method able to analyze various compounds relatively fast (Lindinger et al., 1998). The response times of most of these instruments vary between 0.8 and 1.5 seconds. In the PTR-MS instrument positive ions are formed from target component by proton transfer reaction with primary ion (H3O+). These product ions are then selected with a mass spectrometer and detected with electron multiplier.

35

3.5 Emission modeling

All the methods described above are used to provide values of parameters necessary in emission models for the estimation of regional and global emissions of BVOCs. Emissions fluxes (F) are usually calculated using an expression of the type described by Guenther (1997) F = εDγλ ,

(9)

where ε is the emission potential, or the basal emission rate, D the foliar biomass density,

γ the short-term emission activity factor to account for light and temperature dependencies of emission, and λ the long-term, or seasonal, activity factor.

The short-term activity factor for isoprene is a function of light and temperature and can be represented as:    C (T − Ts )   exp T 1    αC L1 L   RTs T   , γ =  2 2    1 + α L   C + exp CT 2 (T − TM )   T3   RTs T    

(10)

where L is the photosynthetic photon flux density (PPFD), T is the leaf temperature, Ts is the leaf temperature at standard condition (usually 30ºC), R is the gas constant, and CL1,

α, CT1, CT2 and CT3 are empirical constants (Guenther et al., 1991, 1993), with typical values as in, e.g., Guenther (1997). Monoterpene emissions usually depend only on temperature and they are calculated using:

γ = exp[β (T − Ts )] ,

(11)

where β is an empirical coefficient usually taken to be 0.09 ºC-1 (Guenther et al., 1991).

36

However, monoterpene emissions from some Mediterranean and tropical plant species are shown to be also light-dependent (Staudt and Seufert, 1995; Loreto et al., 1996; Kuhn et al., 2001). The total emission from plant species with partially light-dependent monoterpene emissions is modeled by assuming two independent sources, one of which is temperature-dependent pool emission, and the other is light and temperature-dependent synthesis emission, thus: F = F pool (T ) + Fsynth. (T , PPFD ) ,

(12)

where the first term on the right hand-side is calculated using equation (11), and the second term using equation (10).

The long-term activity factor accounts for the seasonal changes in the emission potential. There is no simple algorithm for this factor. Seasonal changes in basal emission rates have been observed for both isoprene and monoterpenes. Part of these changes can be due to the environmental parameters, part due to phenology (e.g. Guenther 1997; Hakola et al., 1998; Hansen and Seufert, 1999; Geron et al., 2000; Boissard et al., 2001; Pétron et al., 2001).

The dependencies of BVOC emissions on various factors within the emission models are described in varying detail depending on the nature of the problem. Overall, regional and global models can be divided into those where species-specific emission parameters are used, and those where ecosystem specific emission parameters are used.

37

The species-specific approach can be used in areas where emissions are dominated by a few plant species, or when the modeling domain is limited. This approach has been used to estimate BVOC emissions on the continental scale by Simpson et al. (1999), and on the regional scale by, e.g., Lindfors and Laurila (2000), Lindfors et al. (2000) and Kellomäki et al. (2001).

In tropical rainforests, on the other hand, the number of tree species per hectare can reach three hundred (Whitmore, 1998, and references therein). Thus, the species-specific approach is highly impractical in these areas. Instead, an ecosystem-specific approach, where different vegetation types are assigned with integrated emission parameters are used. This approach has been implemented for global emission estimates by Guenther et al. (1995).

4 RESULTS

The results of papers I-V show that measurements of hydrocarbon emissions from vegetation are possible using several techniques. However, the uncertainty of a single half-hourly flux value can be large, from about 30 % to over 100 % (Figure 4). This leads to the need of larger data sets. In any technique for trace gas flux measurements the accuracy of the concentration measurement is of vital importance since a small difference in concentration has to be measured.

38

Here the main results of the papers included in the thesis are presented, the more detailed discussion being found in the papers themselves. First, some remarks on the gradient technique (4.1) and disjunct eddy sampling methods (4.2) are made. Results from canopy-scale flux measurement are summarized (4.3), and finally implications for future research are discussed (4.4). JULY 31, 1997

AUGUST 2, 1997

300

300

250

250

200

200

150

150

100

100

50

50 0

0 0

3

6

9

12

15

18

21

24

0

3

6

TIME

9

12

15

18

21

24

TIME

Figure 4. Monoterpene fluxes (ng m-2 s-1) measured by the gradient technique, together with their estimated uncertainties at Huhus, Ilomantsi, Finland (Rinne et al., 1999). The estimation of uncertainties is conducted as described in paper I.

4.1 Gradient technique

The gradient technique using the Monin-Obukhov stability functions proved to be a useful framework for estimating the hydrocarbon emission rates at the canopy scale. However, the data obtained using this method needs to be carefully analyzed to exclude possible unphysical data. The sources of uncertainty were analyzed in paper I. The results show the importance of accurate trace gas sampling and analysis. The accuracy of the trace gas concentration measurements was improved for the measurements presented

39

in paper II mainly by lowering the blank-level variation of the solid adsorbent cartridges through more thorough cleaning procedure.

The analyzed uncertainties can be divided into random and systematic errors. The latter can bias the results of any statistical analysis of the flux measurement results. If the concentration measurements can be assumed to have no systematic error, the most important potential sources of systematic error for the gradient method using the MoninObukhov stability functions are the parameterizations and the effect of the roughness sublayer. These can introduce a systematic error of 30-40 % to the measured fluxes.

4.2 Disjunct eddy sampling

Papers III and V describe new methods for hydrocarbon flux measurements using the

disjunct eddy sampling approach. Simulations show that disjunct eddy sampling does give results similar to continuous sampling. The DES can be applied to various eddy flux measurement methods such as the true eddy accumulation technique (paper III) and the eddy covariance technique (paper V). The disjunct eddy sampling relaxes some of the instrumental requirements imposed by continuous sampling.

Both techniques applying the disjunct eddy sampling were successfully deployed to measure fluxes of biogenic hydrocarbons. Particularly the disjunct eddy covariance technique is a potentially powerful method for trace gas flux measurements, being able to create large data sets and being less prone to systematic errors than, e.g., gradient

40

techniques. However, the uncertainties in the calibration of the analyzer used can introduce systematic uncertainties into fluxes measured. For the PTR-MS, the calibration uncertainty was estimated to be 20 %. Intercomparisons are needed to verify the fluxes measured using DES techniques.

Future applications of DES can include e.g. hyperbolic relaxed eddy accumulation method (Bowling et al., 1999). The DEA can also be applied for aircraft based measurements since a delay of a second of two in the on-line wind data can easily be incorporated with the system.

4.3 Canopy-scale VOC emissions

The boreal Scots pine forests and mixed forest in Finland proved to be monoterpene emitters as expected (paper II). The measured canopy scale monoterpene emissions were temperature-dependent as the leaf level enclosure measurements had predicted (e.g. Janson, 1993) but the observed β-coefficient was 0.15 ºC-1, i.e. somewhat higher than 0.09 ºC-1, which is commonly used in emission models (Guenther et al., 1995; Simpson et al., 1999; Lindfors and Laurila, 2000; Lindfors et al., 2000; Kellomäki et al., 2001) and based on enclosure measurements (Guenther et al., 1991, 1993). This was also the case with the α-pinene fluxes measured at Niwot Ridge (paper III) and the data presented by Schween et al. (1997) and Christensen et al. (2000). The β-coefficient for all daytime data measured at Kenttärova (paper I) is 0.11 ºC-1, which is close to the coefficient commonly used. The reasons for the higher β-coefficients obtained by SLF

41

techniques are not known, but they may include the use of air temperature, rather than leaf temperature, in some of the studies; and the effects of the canopy on the radiation, and therefore on leaf temperature distribution.

The parameters used in the emission models are commonly obtained by enclosure measurements conducted under relatively warm temperatures. The extrapolation of emissions to low temperatures can introduce systematic error into the calculated fluxes, if the used β-coefficient differs from the real temperature dependence. Thus, if the higher βcoefficient observed by canopy scale measurements is not an artifact, the emission models may be overestimating the monoterpene emissions from boreal forests under lower temperatures, which are rather common for the area.

Guenther et al. (1995) used a basal monoterpene emission rate of 2.4 µgC gdw-1 h-1 for boreal and snowy coniferous forests whereas Simpson et al. (1999) used 1.3 µgC gdw-1 h-1 for Scots pine and 0.18 µgC gdw-1 h-1 for birches (Betula spp.). For Norway spruce (Picea abies) Simpson et al. (1999) used partially light-dependent emission algorithm (equation (12)) with basal emission rate of 1.3 µgC gdw-1 h-1 for both pool and synthesis emission. According to the measurements reported in papers I and II, the basal monoterpene emission rates were 1.5 µgC gdw-1 h-1 for mixed northern boreal forest (Siberian spruce/mountain birch: Picea abies subsp. obovata/Betula pubenscens subsp. czerepanovii), and 1.1 µgC gdw-1 h-1 for Scots pine (Pinus sylvestris) forest at the border between the mid- and southern boreal subzones. Christensen et al. (2000) reported a basal monoterpene emission rate for Norway spruce to be 0.64 µgC gdw-1 h-1, obtained by

42

REA method.

Thus, it seems that the emission models have overestimated the

monoterpene emissions from the European boreal zone due to biased basal emission rates. Too high basal monoterpene emission rates used in emission models may be due to the destruction of fine organs storing monoterpenes in the enclosure measurements, leading to artificially high emission rates measured. Also the extrapolation of known emission parameters to related plant species or ecosystems, for which emission parameters are not known, can introduce large errors to model results.

The basal isoprene emission rate used by Guenther et al. (1995) for boreal and snowy coniferous forests is 8 µgC gdw-1 h-1, which results in a relatively high isoprene emissions. Simpson et al. (1999) used basal isoprene emission rate of 1.0 µgC gdw-1 h-1 for Norway spruce. In the northern boreal zone, as well as at the measurement site described in paper I, the dominant spruce is Siberian spruce, which is a sub-species of Norway spruce.

According to Steinbrecher et al. (1999), Siberian spruce has lower basal isoprene emission rates than Norway spruce and Lindfors et al. (2000) used basal isoprene emission rate of 0.1 µgC gdw-1 h-1 for Siberian spruce. Scots pine and birches have generally been regarded as low or non-isoprene emitters (Simpson et al., 1999; Hakola et al., 1998, 1999; Lindfors et al., 2000). The isoprene fluxes at the boreal sites were below the detection limit during the measurements described in papers I and II. The strongest isoprene emitters in the European boreal zone seem to be scrubby plants growing patchily, often along waters and in wetlands, such as willows (Hakola et al., 1998, see also Janson et al., 1999). Therefore, the contribution of these plants to the isoprene emission escapes the measurements presented in papers I and II. The VOC flux

43

measurements conducted in the North American boreal zone have shown relatively high isoprene emissions from aspen and black spruce forests (Zhu et al., 1999; Westberg et al., 2000).

Simpson et al. (1999) used a foliar biomass of 500 g m-2 for Scots pine in the areas north of 60°N, whereas the detailed analysis of the Finnish forest inventory data by Kellomäki et al. (2001) shows that the highest foliar biomass densities for pure Scots pine stands in southern Finland (60-65°N) are 200-290 g m-2. For Norway spruce Simpson et al. (1999) used foliar biomass density of 800 g m-2 in the areas north of 60°N, which falls in the range of 350-950 g m-2 reported by Kellomäki et al. (2001) for southern Finland. In northern Finland the foliar biomasses of Scots pine and Norway spruce are 80-90 % and 40-50 % of the foliar biomasses in southern Finland, respectively (Kellomäki et al., 2001). The recent regional emission inventories for Finland by Lindfors and Laurila (2000), Lindfors et al. (2000) and Kellomäki et al. (2001) use basal emission rates which fit the data in papers I and II, and foliar biomass densities based on recent analysis.

Compared to the results presented in papers I and II and other published data (Christensen et al., 2000), global and continental scale biogenic hydrocarbon emission inventories (e.g., Guenther et al., 1995; Simpson et al., 1999) have generally overestimated the monoterpene emissions from the European boreal zone. This overestimation is due to either too high basal emission rates or too high foliage biomasses used in emission calculations or both and can be as high as 100 %. Isoprene emission is also likely to be overestimated, but the measurement data presented here does not allow

44

quantitative estimate of the magnitude of overestimation. We must, however, remember that the regional and global emission models describe emissions from large areas, whereas SFL measurements provide information on a much smaller scale. For example, the effect of patchily growing scrub vegetation on the spatially averaged VOC emission must be taken into account in the emission models.

400

Flux [ng m-2 s-1]

300

200

100

0 7/7/00 00:00

7/7/00 12:00

7/8/00 00:00

7/8/00 12:00

7/9/00 00:00

Figure 5. Measured isoprene (solid circles), α-pinene (open diamonds) and β–pinene (open triangles) fluxes at Km 67, Floresta Nacional do Tapajós, Pará, Brazil (Paper IV).

In tropical rainforests the isoprene emissions are reported to be higher than monoterpene emissions (Zimmerman et al., 1988; Helmig et al., 1998). This was also the case in measurements reported in paper IV (Figure 5). Even when the hydrocarbon emission from the rainforest is dominated by isoprene, the monoterpene emission is of the same order of magnitude as those measured in the European boreal region. These emissions have previously been modeled as temperature-dependent emissions, in a similar fashion as monoterpene emission from boreal region (Guenther et al., 1995). Measurements

45

conducted at km 67 in Floresta Nacional do Tapajós, Pará, Brazil (paper IV), however, show that these emissions are likely to be temperature- and light-dependent in a manner similar to isoprene emission. The light dependence of the monoterpene fluxes have an effect on the diurnal cycle of monoterpene emissions and on the nighttime air chemistry in the tropical regions, as the monoterpenes do not reach high concentrations in the shallow nocturnal boundary layer, as in the European boreal zone. Since the parameters used in modeling the monoterpene emissions from tropical rainforests are obtained from daytime data, the extrapolation of these parameters to night conditions using algorithms depending solely on temperature can lead to significant overestimation of the monoterpene emissions. For example, if we calculate the emission using the same basal emission rate by different algorithms, we find that the algorithm depending solely on temperature (equation (11)) results in three time higher daily emission than the temperature and light dependent algorithm (equation (10)).

The isoprene fluxes at km 67 did not correlate as well with light and temperature dependent emission activity factor (equation 10) as α-pinene fluxes did, as would be expected. The observed isoprene flux variation, not correlated with light- and temperature algorithm, could be caused by heterogeneities in horizontal source distribution. This could be caused by nearby gaps in the canopy, likely to be occupied by fast growing high isoprene emitters. To average out the effect of the horizontal inhomogeneity, the measurements should have been conducted higher above the canopy than was the case. However, the emission pattern observed, with high isoprene emission and lower monoterpene emission, is similar to those derived from well-mixed boundary layer

46

profiles (Zimmerman et al., 1988; Helmig et al., 1998) and calculated by emission models (e.g. Guenther et al., 1995). As there is no foliar biomass density data for km 67 available yet, a direct comparison of measured fluxes with modeled emission is not possible. As these measurements were conducted at only one site in the hugely diverse neo-tropical forest region, the characterization of the VOC emissions from tropical vegetation clearly requires more measurements.

0.12

0.012

Methanol

0.008

Methanol g m-2

0.08

Acetone & Acetaldehyde g m-2

Cutting started

Acetaldehyde Acetone

0.04

0.004

0

0 12-Aug

13-Aug

14-Aug

15-Aug

Figure 6. Cumulative methanol, acetone and acetaldehyde fluxes from alfalfa before and after cutting in Morgan County, Colorado, USA, August 2000. The data used for the integration is presented by Warneke et al. (2001).

47

Paper V, which concentrates on the methodological development, shows also some new

data on the methanol fluxes from undisturbed and drying alfalfa. Undisturbed alfalfa emitted significant amounts of methanol. As other plant species have also been shown to emit methanol (MacDonald and Fall, 1993; Nemecek-Marshall et al., 1995; Fukui and Doskey, 1998; Kirstine et al., 1998), vegetation can be a more important source of methanol into the atmosphere than previously estimated. As the lifetime of methanol is relatively long, order of 16 days (Singh et al., 1995), it can easily reach the free troposphere and thus affect the chemistry there. Methanol concentrations up to 700 pptv have been observed in the free troposphere (Singh et al., 1995; 2000). Drying alfalfa emitted also acetaldehyde and acetone but their emissions were an order of magnitude lower than methanol emissions (Figure 6). Since the undisturbed alfalfa did not emit these compounds, the effect of these emissions may not be significant at the global scale. However, since acetaldehyde is much more reactive than methanol and acetone, it is more likely to have an impact on the local air chemistry. The cumulative fluxes (Figure 6) also revealed deposition of acetone onto the alfalfa. However, relatively little is still known about the biogenic emissions of oxygenated VOCs, such as methanol.

4.4 Implications for future research

The results show that several surface layer flux methods can be used to obtain canopyscale data on the VOC emissions from vegetation. However, direct eddy covariance measurements using PTR-MS, sampling either continuously or in a disjunct manner, are

48

likely to gain popularity in the future as a state of the art technology. The eddy covariance method is more applicable for acquisition of larger flux data sets than gradient or eddy accumulation methods, which require laborious laboratory analysis of the samples. These larger data sets can be used for process studies and they can reveal more easily weak processes, such as the deposition of acetone onto alfalfa. The DEC method may be also used for measurement of the fluxes of other compounds such as methane and mercury using suitable analyzers. The large uncertainties associated with a single halfhour flux value imply a need for larger datasets to accurately determine the canopy scale VOC emission rates from various types of vegetation. This is especially true for tropical rainforests where characterizing all plant species by enclosure measurements is practically impossible.

Although this work covers only a small part of all vegetation types, the results bear some implications for future emission modeling studies. According to the measurements conducted in Finland, the monoterpene emissions from European boreal forests have been overestimated in many emission models. This calls for revision of basal emission rates and foliar biomass data used in model calculations. The measurements conducted at the Amazonian rainforest showed dependence of the monoterpene emissions on light. Application of light-dependent monoterpene emission algorithm into emission models of neo-tropical rainforests affects the diurnal cycle of the emission and can have an effect on the total emission as well.

49

Some of the findings presented in this work invite for deeper studies in the near future. The extent of the light dependence of monoterpene emissions among tropical plant species is one of these, as tropical forests are a major source of BVOCs into the atmosphere. Little is still known about biogenic emission of oxygenated VOCs such as methanol. If the high methanol emissions observed from alfalfa are common among other plant species, terrestrial vegetation may emit higher amounts of this compound into the atmosphere than expected. The effect of land cover change was not studied within this work but it can be an important issue in understanding the human impact on atmospheric composition. Surface layer flux measurement methods are a good tool for these studies. For the canopy-scale emission measurements, the intercomparisons to validate the flux measurement techniques should be in high priority.

5 SUMMARY AND CONCLUSIONS

The results of this work indicate that: 1) Measurements of biogenic VOC emissions are possible using various surface layer flux techniques. The uncertainties associated with these techniques are relatively large. To ensure data quality, careful analysis of the possible error sources and micrometeorological conditions is needed. 2) Developed new methods based on disjunct eddy sampling approach can provide more and better data on BVOC emission including oxygenated compounds. Especially the disjunct eddy covariance method can be a powerful new tool for trace gas flux measurements.

50

3) Chemical analysis of VOCs was one of the most important sources of uncertainty for surface layer flux measurements of these compounds. 4) The European boreal forest emitted mostly monoterpenes. These emissions were, however, somewhat smaller than previously estimated in emission inventories. The measured canopy-scale monoterpene emissions showed temperaturedependent behavior. 5) In the neo-tropical rainforest, isoprene emission dominated over the monoterpene emissions. Monoterpene emissions showed light and temperature dependent behavior similar to isoprene emission. 6) Both undisturbed and disturbed alfalfa showed high methanol emissions. If these high methanol emissions are common among other vegetation types, they can have a significant effect on the global atmospheric methanol budget. 7) The results presented in this work cover only a small part of the global VOC emissions from the biosphere. They also opened new questions on certain emissions. There is certainly a need for intercomparison studies between different trace gas flux measurement techniques. VOC emissions from rainforests are still not very well characterized and the extent of light dependent monoterpene emissions is unknown. Also emissions from croplands need more research as well as emissions of oxygenated VOCs in general.

51

REFERENCES

Andersson-Sköld, Y. & Simpson, D., 2001: Secondary organic aerosol formation in northern Europe: A model study. J. Geophys. Res., 106, 7357-7374.

Atkinson, R., 2000: Atmospheric chemistry of VOCs and NOx. Atmos. Environ., 34, 2063-2101.

Baker, J. M., Norman, J. M. & Bland, W. L., 1992: Field-scale application of flux measurement by conditional sampling. Agric. For. Meteorol., 62, 31-52.

Baker, B., Guenther, A., Greenberg, J., Goldstein, A. & Fall, R., 1999: Canopy fluxes of 2-methyl-3-buten2-ol over a ponderosa pine forest by relaxed eddy accumulation: Field data and model comparison. J. Geophys. Res., 104, 26107-26114.

Boissard, C., Cao, X.-L., Juan, C.-Y., Hewitt, C. N. & Gallagher, M., 2001: Seasonal variations in VOC emission rates from gorse (Ulex europaeus). Atmos. Environ., 35, 917-927.

Bowling, D. R., Delany, A. C., Turnipseed, A. A., Baldocchi, D. D. & Monson, R. K., 1999: Modification of the relaxed eddy accumulation technique to maximize the measured scalar mixing ratio differences in updrafts and downdrafts. J. Geophys. Res., 104, 9121-9133.

Businger, J. A., Wyngaard, J. C., Izumi, Y. & Bradley, E. F., 1971: Flux-profile relationships in the atmospheric surface layer. J. Atmos. Sci., 28, 181-189.

Businger, J. A., 1986: Evaluation of the accuracy with which dry deposition can be measured with current micrometeorological techniques. J. Clim. Appl. Meteorol., 25, 1100-1124.

52

Businger, J. A., & Delany, A. C., 1990: Chemical sensor resolution required for measuring surface fluxes by three common micrometeorological techniques, J. Atmos. Chem., 10, 399-410.

Businger, J., A., & Oncley, S. P., 1990: Flux measurement with conditional sampling. J. Atmos. Ocean. Tech., 7, 349-352.

Buzorius, G., Rannik, Ü., Mäkelä, J., Vesala, T. & Kulmala, M., 1998: Vertical aerosol particle fluxes measured by eddy covariance technique using condensational particle counter. J. Aerosol Sci., 29, 157-171.

Cao, X.-L., Boissard, C., Juan, A. J., Hewitt, C. N., Gallagher, M., 1997: Biogenic emissions of volatile organic compounds from gorse (Ulex europaeus): Diurnal emission fluxes at Kelling Heath, England. J. Geophys. Res., 102, 18903-18915.

Cao, X.-L. & Hewitt, C. N., 1999: The sampling and analysis of volatile organic compounds in the atmosphere. In: Hewitt, C. N. (Ed.): Reactive hydrocarbons in the Atmosphere. Academic Press,

Cellier, P., & Brunet, Y., 1992: Flux-gradient relationships above tall plant canopies. Agric. For. Meteorol., 58, 93-117.

Chameides, W. L., Fehsenfeld, F., Rogers, M., Cardelino, C., Martinez, J., Parrish, D., Lonneman, W., Lawson, D. R., Rasmussen, R. A., Zimmerman, P., Greenberg, J., Middleton, P. & Wang, T., 1992: Ozone precursor relationships in the ambient atmosphere. J. Geophys. Res., 97, 6037-6055.

Christensen, C. S., Hummelshøj, P., Jensen, N. O., Larsen, B., Lohse, C., Pilegaard, K. & Skov, H., 2000: Determination of the terpene flux from orange species and Norway spruce by relaxed eddy accumulation. Atmos. Environ., 34, 3057-3067.

53

Ciccioli, P., Brancaleoni, E., Frattoni, M., Di Palo, V., Valentini, R., Tirone, G., Seufert, G., Bertin, N., Hansen, U., Csiky, O., Lenz, R. & Sharma, M., 1999: Emission of reactive terpene compounds from orange orchards and their removal by within-canopy processes. J. Geophys. Res., 104, 8077-8094.

Cordova Leal, A. M., Vanni Gatti, L., Yamazaki, A., Silva Dias, M. A., & Artaxo, P., 2000: A special case: High concentrations of ozone in the nighttime due to effects of a convective system. Book of abstracts of First LBA Scientific Conference, June 26-30, 2000, Belém, Pará, Brazil.

Crutzen, P. J. & Zimmerman, P. H., 1991: The changing photochemistry of the troposphere. Tellus, 43, 136-151.

Dabberdt, W. F., Lenschow, D. H., Horst, T. W., Zimmerman, P. R., Oncley, S. P. & Delany, A. C., 1993: Atmosphere-surface exchange measurements. Science, 260, 1472-1481.

Darmais, S., Dutaur, L., Larsen, B., Cieslik, S., Luchetta, L., Simon, V. & Torres, L., 2000: Emission fluxes of VOC by orange trees determined by both relaxed eddy accumulation and vertical gradient approaches. Chemosphere, Global Change Science, 2, 47-56.

Desjardins, R. L., 1977: Description and evaluation of sensible heat flux detector. Boundary-Layer Meteorol., 11, 147-154.

Dyer, A. J., 1974: A review of the flux profile relations. Boundary-Layer Meteorol., 1, 363-372.

Evans, R. C., Tingey, D. T. & Gumpertz, M. L., 1985: Interspecies variation in terpenoid emissions from Engelmann and Sitka spruce seedlings. Forest Sci., 31, 132-142.

54

Fehsenfeld, F., Calvert, J., Fall, R., Goldan, P., Guenther, A., Hewitt, N., Lamb, B., Liu, S., Trainer, M., Westberg, H. & Zimmerman, P., 1992: Emissions of volatile organic compouds from vegetation and the implications for atmospheric chemistry. Global Biogeochemical Cycles, 6, 389-430.

Fuentes, J. D., Wang, D., Neumann, H. H., Gillespie, T. J., Den Hartog, G., & Dann, T. F., 1996: Ambient biogenic hydrocarbons and isoprene emissions from a mixed deciduous forest. J. Atmos. Chem., 25, 67-95.

Fuentes, J. D., Lerdau, M., Atkinson, R., Baldocchi, D., Bottenheim, J. W., Ciccioli, P., Lamb, B., Geron, C., Gu, L., Guenther, A., Sharkey, T. D. & Stockwell, W., 2000: Biogenic hydrocarbons in the atmospheric boundary layer: A review. Bull. Am. Meteor. Soc., 81, 1537-1575.

Fukui, Y. & Doskey, P. V., 1998: Air-surface exchange of nonmethane organic compounds at a grassland site: Seasonal variations and stressed emissions. J. Geophys. Res., 103, 13153-13168.

Gallagher, M. W., Clayborough, R., Beswick, K. M., Hewitt, C N., Owen, S., Moncrieff, J. & Pilegaard, K., 2000: Assessment of a relaxed eddy accumulation for measurements of fluxes of biogenic volatile organic compounds: Study over arable crops and a mature beech forest. Atmos. Environ., 34, 2887-2899.

Garratt, J. R., 1980: Surface influence upon vertical profiles in the atmospheric near-surface layer. Q. J. R. Meteorol. Soc., 106, 803-819.

Garratt, J. R., 1994: The atmospheric boundary layer. First paperback edition with corrections. Cambridge University Press, Cambridge, UK. pp. 316.

Geron, C., Guenther, A., Sharkey, T., & Arnts, R. R., 2000: Temporal variation in basal isoprene emission factor. Tree Physiology, 20, 799-805.

55

Goldstein, A. H., Fan, S. M., Goulden, M. L., Munger, J. W. & Wofsy, S. C., 1996: Emissions of ethane, propene and 1-butene by a midlatitude forest. J. Geophys. Res., 101, 9149-9157.

Granier, C., Pétron, G., Müller, J.-F., & Brasseur, G., 2000: The impact of natural and anthropogenic hydrocarbons on the tropospheric budget of carbon monoxide. Atmos. Environ., 34, 5255-5270.

Greenberg, J. P., Guenther, A. B., Madronich, S., Baugh, W., Ginoux, P., Druilhet, A., Delmas, R. & Delon, C., 1999a: Biogenic volatile organic compound emissions in central Africa during the Experiment for the Sources and Sinks of Oxidants (EXPRESSO) biomass burning season. J. Geophys. Res., 104, 30659-30671.

Greenberg, J. P., Guenther, A., Zimmerman, P., Baugh, W., Geron, C., Davis, K., Helmig, D. & Klinger, L. F., 1999b: Tethered balloon measurements of biogenic VOCs in the atmospheric boundary layer. Atmos. Environ., 33, 855-867.

Griffin, R. J., Cocker, D. R. III, Seinfeld, J. H. & Dabdub, D., 1999: Estimate of global atmospheric organic aerosol from oxidation of biogenic hydrocarbons. Geophys. Res. Lett., 26, 2721-2724.

Guenther, A. B., Monson, R. K., & Fall, R., 1991: Isoprene and monoterpene emission rate variability: Observations with eucalyptus and emission rate algorithm development. J. Geophys. Res., 96, 1079910808.

Guenther, A., Zimmerman, P., Harley, P., Monson, R., Fall, R., 1993: Isoprene and monoterpene emission rate variability: Model evaluation and sensitivity analysis. J. Geophys. Res., 98, 12609-12617.

Guenther, A., Hewitt, C. N., Erickson, D., Fall, R., Geron, C., Graedel, T., Harley, P., Klinger, L., Lerdau, M., McKay W. A., Pierce T., Scholes, B., Steinbrecher, R., Tallamraju, R., Taylor, J. & Zimmerman, P., 1995: A global model of natural volatile organic compound emissions. J. Geophys. Res., 100, 8873-8892.

56

Guenther, A., Baugh, W., Davis, K., Hampton, G., Harley, P., Klinger, L., Vierling, L., Zimmerman, P., Allwine, E., Dilts, S., Lamb, B., Westberg, H., Baldocchi, D., Geron, C. & Pierce, T., 1996: Isoprene fluxes measured by enclosure, relaxed eddy accumulation, surface layer gradient, mixed layer gradient, and mixed layer mass balance techniques. J. Geophys. Res., 100, 18555-18567.

Guenther, A., 1997: Seasonal and spatial variations in natural volatile organic compound emissions. Ecological Applications, 7, 34-45.

Guenther, A. B., & Hills, A. J., 1998: Eddy covariance measurement of isoprene fluxes. J. Geophys. Res., 103, 13145-13152.

Guenther, A., Archer, S., Greenberg, J., Harley, P., Helmig, D., Klinger, L., Vierling, L., Wildermuth, M., Zimmerman, P. & Zitzer, S., 1999: Biogenic hydrocarbon emission and landcover/climate change in a subtropical savanna. Phys. Chem. Earth, 24, 659-667.

Hakola, H., Arey, J., Aschmann, S. M. & Atkinson, R., 1994: Product formation from the gas-phase reactions of OH radicals and O3 with a series of monoterpenes. J. Atmos. Chem., 18, 75-102.

Hakola, H., Rinne, J. & Laurila, T., 1998: Hydrocarbon emission rates of tea-leafed willow (Salix phylicifolia), Silver birch (Betula pendula) and European aspen (Populus tremula). Atmos. Environ., 32, 1825-1833.

Hakola, H., Rinne, J. & Laurila, T., 1999: The VOC emission rates of boreal deciduous trees. In: Laurila, T., and V. Lindfors (eds.) Biogenic VOC emissions and photochemistry in the boreal regions of Europe. Air Pollution Research Report 70, Commission of European Communities, Luxembourg, pp. 21-28.

57

Hakola, H., Laurila, T., Rinne, J. & Puhto, K., 2000: The ambient concentrations of biogenic hydrocarbons at a Northern European, boreal site. Atmos. Environ., 34, 4971-4982.

Hansen, U. & Seufert, G., 1999: Terpenoid emission from Citrus sinensis (L.) OSBECK under drought stress. Phys. Chem. Earth, 24, 681-687.

Haugen, D. A., 1978: Effects of sampling rates and averaging periods on meteorological measurements. Proc. Fourth Symp. On Meteorological Observations and Instrumentation, Denver, CO. Amer. Meteor. Soc., 15-18.

Helmig, D., Balsley, B., Davis, K., Kuck, L. R., Jensen, M., Bognar, J., Smith Jr, T., Vasquez Arrieta, R., Rodríguez, R. & Birks, J. W., 1998: Vertical profiling and determination of landscape fluxes of biogenic nonmethane hydrocarbons within the boundary layer in the Peruvian Amazon. J. Geophys. Res., 103, 25519-25532.

Hoffmann, T., Odum, J. R., Bowman, F., Collins, D., Klockow, D., Flagan, R. F. & Seinfeld, J. H., 1997: Formation of organic aerosols from the oxidation of biogenic hydrocarbons. J. Atmos. Chem., 26, 189-222.

Horst, T. W., 1999: The footprint estimation of atmosphere-surface exchange fluxes by profile techniques. Boundary-Layer Meteorol., 90, 171-188.

IPCC, 1996: Climate Change 1995, The Science of Climate Change. Cambridge University Press, Cambridge, UK. pp. 572.

Isidorov, V. A., Zenkevich, I. G. & Ioffe, B. V., 1985: Volatile organic compounds in the atmosphere of forest. Atmos. Environ., 19, 1-8.

58

Janson, R., 1993: Monoterpene emissions from Scots pine and Norwegian spruce. J. Geophys. Res., 98, 2839-2850.

Janson, R. W., De Serves, C. & Romero, R., 1999: Emission of isoprene and carbonyl compounds from a boreal forest and wetland in Sweden. Agric. For. Meteorol., 98-99, 671-681.

Juuti, S., Arey, J. & Atkinson, R., 1990: Monoterpene emission rate measurements from a Monterey pine. J. Geophys. Res., 95, 7515-7519.

Kaimal, J. C. & Gaynor, J. E., 1983: The Boulder Atmospheric Observatory. J. Clim. Appl. Meteor., 22, 863-880.

Kaimal, J. C. & Finnigan, J. J., 1994: Atmospheric Boudary Layer Flows. Their Structure and Measurement. Oxford University Press, New York, USA. pp. 289.

Karl, T., Guenther, A., Jordan, A., Fall, R. & Lindinger, W., 2001: Eddy covariance measurement of biogenic oxygenated VOC emissions from hay harvesting. Atmos. Environ., 35, 491-495.

Kavouras, I. G., Mihalopoulos, N. & Stephanou, E. G., 1998: Formation of atmospheric particles from organic acids produced by forests. Nature, 395, 683-686.

Kellomäki, S., Rouvinen, I., Peltola, H. & Strandman, H., 2001: Density of foliage mass and area in the boreal forest cover in Finland, with applications to the estimation of monoterpene and isoprene emissions. Atmos. Environ., 35, 1491-1503.

Kesselmeier, J., Kuhn, U., Wulf, A., Andreae, M. O., Ciccioli, P., Brancaleoni, E., Frattoni, M., Guenther, A., Greenberg, J., de Castro Vasconcellos, P., De Oliva, T., Tavares, T. & Artaxo, P., 2000: Atmospheric

59

volatile organic compounds (VOC) at a remote tropical forest site in central Amazonia. Atmos. Environ., 34, 4063-4072.

Kirstine, W., Galbally, I., Yuerong, Y., Hooper, M., 1998: Emissions of volatile organic compounds (primarily oxygenated species) from pasture. J. Geophys. Res., 103, 10605-10619.

Klinger, L. F., Greenberg, J., Guenther, A., Tyndall, G., Zimmerman, P., M’Bangui, M., Moutsaboté, J.-M. & Kenfack, D., 1998: Patterns of volatile organic compound emissions along a savanna-rainforest gradient in central Africa. J. Geophys. Res., 103, 1443-1454.

Kristensen, L., Mann, J., Oncley, S. P. & Wyngaard, J. C., 1997: How close is close enough when measuring scalar fluxes with displaced sensors? J. Atmos. Oceanic Technol., 14, 814-821.

Kuhn, U., Rottenberger, S., Biesenthal, T., Wolf, A., Schebeske, G., Ciccioli, P., Brancaleoni, E., Frattoni, M., Tavares, T. M., & Kesselmeier, J., 2001: Isoprene and monoterpene emissions of Amazonian tree species during the wet season: Direct and indirect investigations on controlling environmental functions. Submitted for publication in J. Geophys Res.

Kulmala, M., Toivonen, A., Mäkelä, J. M. & Laaksonen, A., 1998: Analysis of the growth of nucleation mode particles observed in Boreal forest. Tellus, 50B, 449-462.

Kulmala, M., Hämeri, K., Mäkelä, J. M., Aalto, P. P., Pirjola, L., Väkevä, M., Nilson, E. D., Koponen, I. K., Buzorius, G., Keronen, P., Rannik, Ü., Laakso, L., Vesala, T., Bigg, K., Seidl, W., Forkel, R., Hoffmann, T., Spanke, J., Janson, R., Shimmo, M., Hansson, H.-C., O’Dowd, C., Becker, E., Paatero, J., Teinilä, K., Hillamo, R., Viisanen, Y., Laaksonen, A., Swietlicki, E., Salm, J., Hari, P., Altimir, N. & Weber, R., 2000: Biogenic aerosol formation in the boreal forest. Boreal Environ. Res., 5, 279-297.

60

Lamb, B., Westberg, H., Allwine, G. & Quarles, T., 1985: Biogenic emissions from deciduous and coniferous trees in the United States. J. Geophys. Res., 90, 2380-2390.

Langenheim, J. H., 1994: Higher plant terpenoids: A phytocentric overview of their ecological roles, J. Chem. Ecol., 20, 1223-1280.

Laurila, T. & Hakola, H., 1996: Seasonal cycle of C2-C5 hydrocarbons over the Baltic Sea and Northern Finland. Atmos. Environ., 30, 1597-1607.

Laurila, T. & Lindfors, V. (eds.), 1999: Biogenic VOC emissions and photochemistry in the boreal regions of Europe, Air pollution research report No 70, Commission of the European Communities, Luxembourg, pp. 158.

Lenschow, D. H., Mann, J. & Kristensen, L., 1994: How long is long enough when measuring fluxes and other turbulence statistics? J. Atmos. Ocean. Tech., 11, 661-673.

Lerdau, M., Guenther, A. & Monson, R., 1997: Plant production and emission of volatile organic compounds. BioScience, 47, 373-383.

Lindfors, V. & Laurila, T., 2000: Biogenic VOC emissions from forests in Finland. Boreal Environ. Res., 5, 95-113.

Lindfors, V., Laurila, T., Hakola, H., Steinbrecher, R. & Rinne, J., 2000: Modeling speciated terpenoid emissions from the European boreal forest. Atmos. Environ., 34, 4983-4996.

Lindinger, W., Hansel, A. & Jordan, A., 1998: Proton-transfer-reaction mass spectrometry (PTR-MS): online monitoring of volatile organic compounds at pptv levels. Chem. Soc. Rev., 27, 347-354.

61

Litvak, M. E., Madronich, S. & Monson, R. K., 1999: Herbivore-induced monoterpene emissions from coniferous forests: Potential impact on local tropospheric chemistry. Ecological Applications, 9, 11471159.

Logan, B. A., Monson, R. K. & Potosnak, M. J., 2000: Biochemistry and physiology of foliar isoprene production. Trends in Plant Sciences, 5, 477-481.

Loreto, F., Ciccioli, P., Cecinato, A., Brancaleoni, E., Frattoni, M., Fabozzi, C. & Tricoli, D., 1996: Evidence of the photosynthetic origin of monoterpenes emitted by Quercus ilex L. leaves by 13C labeling. Plant Physiol., 110, 1317-1322.

MacDonald, R. C. & Fall, R., 1993: Detection of substantial emissions of methanol from plants to the atmosphere. Atmos. Environ., 27A, 1709-1713.

Massman, W. J., 2000: A simple method for estimating frequency response corrections for eddy covariance systems. Agric. For. Meteorol., 104, 185-198.

McMillen, R. T., 1988: An eddy correlation technique with extended applicability to non-simple terrain. Boundary-Layer Meteorol., 43, 231-245.

Moore, C. J., 1986: Frequency response corrections for eddy correlation systems. Boundary-Layer Meteorol., 37, 17-35.

Müller, J.-F., 1992: Geographical distribution and seasonal variation of surface emissions and deposition velocities of atmospheric trace gases, J. Geophys. Res., 97, 3787-3804.

Nemecek-Marshall, M., MacDonald, R. C., Franzen, J. J., Wojciechowski, C. L. & Fall, R., 1995: Methanol emission from leaves. Plant Physiol., 108, 1359-1368.

62

Nozière, B., Barnes, I., & Becker, K. H., 1999: Product study and mechanisms of the reactions of α-pinene and pinonaldehyde with OH radicals, J. Geophys. Res., 104, 23645-23656.

NRC, 1991: National Research Council: Rethinking the ozone problem in urban and regional air pollution. National Academy Press, Washington D.C. pp 500.

Paw U, K. T., Baldocchi, D. D., Meyers, T. P. & Wilson, K. B., 2000: Correction of eddy-covariance measurements incorporating both advective effects and density fluxes. Boundary-Layer Meteorol., 97, 487511.

Pétron, G., Harley, P., Greenberg, J. & Guenther, A., 2001: Seasonal temperature variations influence isoprene emission. Geophys. Res. Lett., 28, 1707-1710.

Rannik, Ü., Aubinet, M., Kurbanmuradov, O., Sabelfeld, K. K., Markkanen, T. & Vesala, T., 2000: Footprint analysis for measurements over a heterogeneous forest. Boundary-Layer Meteorol., 97, 137-166.

Rasmussen, R. A., 1972: What do the hydrocarbons from trees contribute to air pollution? J. Air Poll. Control Assoc., 22, 537-543.

Rasmussen, R. A., 1981: A review of the natural hydrocarbon issue. In Bufalini, J. L. and Arnts, R. R. (Eds.): Atmospheric Biogenic Hydrocarbons, vol. 1. Ann Arbor Science Publishers, Ann Arbor, Michigan, USA. 3-14.

Rinne, J., Hakola, H. & Laurila, T., 1999: Vertical fluxes of monoterpenes above a Scots pine stand in the boreal vegetation zone. Phys. Chem. Earth, 24, 711-716.

63

Schaab, G., Steinbrecher, R., Lacaze, B. & Lenz, R., 2000: Assessment of long-term vegetation changes on potential isoprenoid emission for a Mediterranean-type ecosystem in France. J. Geophys. Res., 105, 2886328873.

Schmid, H. P., 1994: Source areas for scalar and scalar fluxes. Boundary-Layer Meteorol., 67, 293-318.

Schuepp, P. H., Leclerc, M. Y., MacPherson, J. I., & Desjardins, R. L., 1990: Footprint prediction of scalar fluxes from analytical solution of the diffusion equation. Boundary-Layer Meteorol., 50, 355-373.

Schween, J. H., Dlugi, R., Hewitt, C. N., & Foster, P., 1997: Determination and accuracy of VOC-fluxes above the pine/oak forest at Castelporziano. Atmos. Environ., 31, 199-215.

Seinfeld, J. H. & Pandis, S. N., 1998: Atmospheric Chemistry and Physics. From Air Pollution to Climate Change. John Wiley & Sons Inc., New York, USA, pp. 1326.

Sharkey, T. D. & Singsaas, E. L., 1995: Why plants emit isoprene. Nature, 374, 769.

Simpson, I. J., Thurtell, G. W., Neumann, H. H., Den Hartog, G., Edwards, G. C., 1998: The validity of similarity theory in the roughness sublayer above forests. Boundary-Layer Meteorol., 87, 69-99.

Simpson, D., Winiwarter, W., Börjesson, G., Cinderby, S., Ferreiro, A., Guenther, A., Hewitt, N., Janson, R., Khalil, M.A.K., Owen, S., Pierce, T., Puxbaum, H., Shearer, M., Skiba, U., Steinbrecher, R., Tarrasón, L., & Öquist, M.G., 1999: Inventorying emissions from nature in Europe. J. Geophys. Res., 104, 81138152.

Singh, H., Chen, Y., Tabazadeh, A., Fukui, Y., Bey, I., Yantosca, R., Jacob, D., Arnold, F., Wohlfrom, K., Atlas, E., Flocke, F., Blake, D., Blake, N., Heikes, B., Snow, J., Talbot, R., Gregory, G., Sachse, G., Vay, S.

64

& Kondo, Y., 2000: Distribution and fate of selected oxygenated organic species in the troposphere and lower stratosphere over the Atlantic. J. Geophys. Res., 105, 3795-3805.

Singh, H. B., Kanakidou, M., Crutzen, P. J. & Jacob, D. J., 1995: High concentrations and photochemical fate of oxygenated hydrocarbons in the global troposphere. Nature, 378, 50-54.

Staudt, M. & Seufert, G., 1995: Light-dependent emission of monoterpenes by holm oak (Quercus ilex L.). Naturwissenschafen, 82, 89-92.

Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic Publishers, Dortrecht, The Netherlands. pp. 666.

Steinbrecher, R., Hauff, K., Hakola, H., Rössler, J., 1999: A revised parameterisation for emission modelind of isoprenoids. In: Laurila, T., and V. Lindfors (eds.) Biogenic VOC emissions and photochemistry in the boreal regions of Europe. Air Pollution Research Report 70, Commission of European Communities, Luxembourg, pp. 29-43.

Tingey, D. T., Manning, M., Grothaus, L. C. & Burns, W. F., 1979: The influence of light and temperature on isoprene emission rates from live oak. Physiol. Plant., 47, 112-118.

Tingey, D. T., Turner, D. P. & Weber, J. A., 1991: Factors controlling the emissions of monoterpenes and other volatile organic compounds. In Sharkey, T. D., Holland, E. A., and Mooney, H. A. (Eds.): Trace gas emissions by plants. Academic Press, San Diego CA, USA. 121-133.

Valentini, R., Greco, S., Seufert, G., Bertin, N., Ciccioli, P., Cecinato, A., Brancaleoni, E. & Frattoni, M., 1997: Fluxes of biogenic VOC from Mediterranean vegetation by trap enrichment relaxed eddy accumulation. Atmos. Environ., 31, 229-238.

65

Vanni Gatti, L., Cordova, A. M., Yamazaki, A., Vasconcellos, M. E., Artaxo, P., Silva Dias, M. A. F., Meixner, F. X., Guenther, A., Bonelli, N., & Aquito, C. A. B., 2000: Dry and wet season measurement of trace gases and aerosols in the Abracos pasture site, Rondonia. Book of abstracts of First LBA Scientific Conference, June 26-30, 2000, Belém, Pará, Brazil.

Vilà-Guerau de Arellano, J., Duynkerke, P. G. & Zeller, K. F., 1995: Atmospheric similarity theory applied to chemically reactive species. J. Geophys. Res., 100, 1397-1408.

Warneke, C., Luxembourg, S. L., de Gouw, J. A., Rinne, H. J. I., Guenther, A. B. & Fall, R., 2001: Disjunct eddy covariance measurements of oxygenated VOC fluxes from an alfalfa field before and after cutting. Submitted for publication in J. Geophys. Res.

Webb, E. K., Pearman, G. I. & Leuning, R., 1980: Correction of flux measurements for density effects due to heat and water vapour transport. Quart. J R. Met. Soc., 106, 85-100.

Went, F. W., 1955: Air pollution. Sci. Am., 192(5), 62-72.

Went, F. W., 1960: Blue hazes in the atmosphere. Nature, 187, 641-643.

Wesely, M. L., & Hart, R. L., 1985: Variability of short term eddy-correlation estimates of mass exchange. In: Hutchison, B., & Hicks, B. (Eds.): The Forest-Atmosphere Interaction. D. Reidel Publishing Company, Dortrecht, The Netherlands, pp. 591-612.

Westberg, H., Lamb, B., Kempf, K. & Allwine, G., 2000: Isoprene emission inventory for the BOREAS southern study area. Tree Phys., 20, 735-743.

Whitmore, T. C., 1998: An Introduction to Tropical Rain Forests. Oxford University Press, Oxford, UK. pp. 282.

66

Wyngaard, J. C. & Brost, R. A., 1984: Top-down and bottom up diffusion in the convective boundary layer. J. Atmos. Sci., 41, 102-112.

Zhu, T., Wang, D., Desjardins, R. L. & Macpherson, J. I., 1999: Aircraft-based volatile organic compounds flux measurements with relaxed eddy accumulation. Atmos. Environ., 33, 1969-1979.

Zimmerman, P. R., Greenberg, J. P. & Westberg, C. A., 1988: Measurements of atmospheric hydrocarbons and biogenic emission fluxes in the Amazon boundary layer, J. Geophys. Res., 93, 1407-1416.

67

APPENDIX A Derivation of the gradient flux equation using Monin-Obukhov stability correction functions

This derivation follows the one given by Fuentes et al. (1996). Garratt (1994) describes derivation of wind, temperature and humidity profiles in thermally stratified surface layer. According to Monin-Obukhov similarity theory, vertical gradient of a mean concentration c in the atmospheric surface layer can be written as −

w' c' ∂c = φ h ( Lz ) , ∂z α cκzu ∗

(A1)

where φh is a dimensionless stability correction function for heat, L is the Obukhov length and κ is the von Kármán constant. If the similarity between scalar transport is assumed, the turbulent Schmidt number αc=1. In gradient measurements described in papers I-II concentrations were measured at two heights. Integrating (A1) between two heights z1 and z2 yields c ( z 2 ) − c ( z1 ) = −

 w' c'   z 2  ln  − Ψh (ζ 2 ) + Ψh (ζ 1 ) , κu∗   z1  

(A2)

where ζ1=z1/L and ζ2=z2/L are the Monin-Obukhov stability parameters at heights z1 and z2, and Ψh is the integral form of the dimensionless stability function, which can be written as: Ψh (ζ ) = ∫ [1 − φ (ζ )]d ln ζ .

Thus the flux Fc is obtained from:

(A3)

68

Fc =

[

− κu ∗ c ( z 2 ) − c ( z1 )

]

   z2  ln  − Ψh (ζ 2 ) + Ψh (ζ 1 )    z1 

.

(A4)

The heights are heights above the zero plane displacement height. In the work described in papers I and II, the integral forms of Businger-Dyer formulas (Businger et al., 1971; Dyer, 1974) were used for calculating the values of the stability correction function Ψh (For these integral forms, see, e.g., Garratt, 1994).

69

APPENDIX B Derivation of the true eddy accumulation equation

Inverting the Reynolds average, the time-averaged covariance of w’ and c’ can take the form: t

1 2 1 w' c' = w' c' dt = ∫ t 2 − t1 t1 t 2 − t1

t2 t 2   ∫ w' cdt − ∫ w' cdt  .  t1  t1

(B1)

As the time average of w’ is zero by definition, the second term on the right hand side of (B1) vanishes. The resulting integral can be divided into updraft and downdraft parts t t t2  1 2 1 2 w ' cdt = w ' c δ dt + w ' c δ dt  , + − ∫t t 2 − t1 t∫1 t 2 − t1  ∫t1 1 

(B2)

where the δ+=1 and δ-=0 when w’>0; and δ+=0 and δ-=1 when w’