Journal of Experimental Botany, Vol. 48, No. 310, pp. 1133-1142, May 1997 Journal of Experimental Botany Trace gas exchange over terrestrial ecosystems: methods and perspectives in micrometeorology John Moncrieff1'5, Riccardo Valentini2, Susanna Greco2, Guenther Seufert3 and Paolo Ciccioli4 1 1nstitute of Ecology and Resource Management, The University of Edinburgh, Mayfield Road, Edinburgh EH93JU, UK 2 University of Tuscia, Dipartimento di Scienze dell'Ambiente Forestale e delle sue Risorse (Di.S.A.F.Ri), Viterbo, Italy 3 Commission of the European Communities, Joint Research Centre (JRC), Environment Institute, Ispra (VA), Italy CNR-lnstitute of Atmospheric Pollution, Montelibretti, Italy Received 15 February 1997; Accepted 25 February 1997 Provided by the author for non-commercial research and education use. Not for reproduction, distribution or commercial use Abstract Quantification of the surface-atmosphere exchange of trace gases is recognized as an essential prerequisite to understanding the role of the biosphere in the global climate system. Among the micrometeorological methods available to measure surface-atmosphere fluxes, the aerodynamic gradient, the energy balance/Bowen ratio, the eddy covariance and the eddy accumulation methods are the most widely employed. This brief review describes the theoretical background and the practical applications of these methodologies and is particularly directed to plant ecophysiologists, ecologists and botanists who may be interested in scaling biological processes to the canopy level. Key words: Trace gas exchange, biosphere, surface-atmosphere fluxes, aerodynamic gradient, Bowen ratio, eddy covariance, eddy accumulation, micrometeorology. Introduction The major biogeochemical cycles on Earth play an important role in the climate system and the role of vegetation in the cycling of carbon, nitrogen and water, at a range of scales, is an area of major research interest. The unresolved budget for global carbon and the location of the 'missing sink' for carbon is a prime example of the progress which needs to be made (Schimel, 1995). ! Similarly, the emission of volatile organic compounds such as the monoterpenes, isoprene etc., which are linked to plant metabolism and photosynthesis, needs to be investigated at the canopy scale to improve our understanding of the potential role of these substances on tropospheric ozone formation. Scaling the functional processes of plants, from cells and organs to communities, is important if such biosphere-atmosphere interaction is to be understood and a quantification of trace gas exchange is essential to validate models at regional and global scale. Using micrometeorological methods to investigate the surface-atmosphere exchange of trace gases is valuable in that they produce fluxes which are integrated over a relatively large area (typically a few km2). The methods are also non-destructive and the results can be used as independent tests of process-based models (Baldocchi et al., 1988). Many reviews have been published on micrometeorological methods in the past (Fowler and Duyzer, 1989; Verma, 1990; Lenschow, 1995), but the pace of change, particularly brought about by developments in available technology, means that scientists not working directly in this field, need to be brought up-todate regularly. In addition, a number of recent papers have clarified methodological issues such as, what is an appropriate sampling period?; at what height should measurements be made?; and how much 'fetch' is enough? (Horst and Weil, 1994; Lenshow et al., 1994). This paper is neither a comprehensive review of all aspects related to measurements of gaseous exchange nor a full list of the techniques available today, but will address the most To whom correspondence should be addressed: Fax: +44 131 662 0478. E-mail: j.moncneff©ed.ac.uk. © Oxford University Press 1997 Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 4 1134 Moncrieff relevant topics which may be used by plant ecologists, botanists and ecophysiologists interested in scaling biological processes to canopy level. Most attention will be focused on the developments and opportunities provided by the direct measuring techniques of eddy covariance and eddy accumulation. The scope of micrometeorology Micrometeorological methods In essence, micrometeorological methods used to quantify turbulent exchange fall into two categories. There are direct methods which sample the air as it flows past a sampling point for its vertical wind speed and direction and its gas concentration (the eddy covariance or eddy accumulation methods) or there are indirect methods based on quantifying the rate of diffusion down a concen- Methods based on concepts of turbulent diffusion Aerodynamic technique: In this method, the transport process is described in a flux-gradient format and one which defines the constant of proportionality K, known as an eddy diffusivity. Thus, the flux, Fx of a trace gas with concentration x, in this form would be: The exchange coefficient (Kx) is obtained by invoking a principle of similarity which says that all scalars (heat, mass) are transported equally effectively, and, once the exchange coefficient for one is found, the rest are equal. In practice, the exchange coefficient for momentum is obtained by anemometry and, via a series of empirically determined steps, the exchange coefficient for the species of interest is obtained. The method has been described fully in Thorn (1975) and Monteith and Unsworth (1990). Fowler and Unsworth (1979) showed how the method could be used to measure fluxes of sulphur dioxide to a cereal crop. Sutton et al. (1993) show the steps involved in calculating fluxes of ammonia using this method, including allowing for the influence of atmospheric stability. In general, the method is difficult to apply when gradients are small, as they often are near to the canopy in tall, rough vegetation such as forests and it cannot easily be used within the roughness sub-layer. It cannot Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 Micrometeorology is the study of the interaction of the lower atmosphere with the surface of Earth and, in effect, it deals with the exchange processes of radiation, mass and momentum which dictate the behaviour of much of what is regarded as weather. Just as micrometeorology is the study of the interface between the atmosphere and the surface, so micrometeorology can be regarded as an interface between the academic subjects of biology and physics (at least fluid dynamics). The atmospheric boundary layer most frequently examined in traditional groundor tower-based micrometeorology is called the surface layer and is usually about 10% the depth of the Atmospheric Boundary Layer (itself denned by Lenschow (1995) as 'the lower part of the atmosphere that interacts with the biosphere and is closely coupled to the surface by turbulent exchange processes'). In the surface layer, fluxes remain within about 10% of their surface value and it is a region where the influence of the coriolis force can be considered negligible. Within the surface layer, two sub-layers can be defined. The roughness sub-layer is a region in which the vegetation elements directly influence the nature of atmospheric turbulence through phenomena such as wake turbulence or thermals which augment and enhance turbulent exchange (Raupach and Thorn, 1981). The depth of the roughness sub-layer can be difficult to estimate, for example, it may extend up to a height of about 3>hc where hc is the height of the canopy (Kaimal and Finnegan, 1994) or its depth may be defined as being within about three times the spacing of the horizontal elements of the canopy (Lenschow, 1995). The layer above the roughness sub-layer is known as the inertia! sub-layer in which fluxes are more or less constant with height and in neutral conditions the wind profile is logarithmic with height (Thorn, 1975). Not all micrometeorological methods are appropriate in all these layers, nor within plant canopies as will be discussed. tration gradient (the aerodynamic and Bowen ratio methods). Each method has distinct advantages and disadvantages and all have been well validated over the course of the past three decades. The direct methods involve sampling at one height only, but with relatively sophisticated sensors and logging equipment. The methods based on measuring gradients obviously require measurements at two or more heights, but use simpler sensors. The disadvantage of the gradient technique is, however, that a number of empirical functions may be required to account for thermal stratification of the atmosphere; additionally, the gradients in atmospheric properties become very small close to vegetation canopies. All the above techniques, when used above vegetation, require that steady-state conditions exist, i.e. that atmospheric conditions are not changing rapidly over the sampling period; they also all require extensive upwind areas to be homogenous, i.e. these methods cannot be used on isolated plots or small fields. If these conditions are met, then it is assumed that the flux measured just above the vegetation is equal to that at the ground or plant surfaces, and fluxes are constant with height, up to a level dependent on the extent of upwind surface homogeneity and atmospheric mixing. These methods will be described briefly in turn and some accompanying methodological issues mentioned. Trace gas exchange be used within plant canopies because of the inapplicability of gradient diffusion theory in this region (Raupach, 1988). The method does have the advantage of using simple instrumentation. Bowen ratio-energy balance method: Following the energy conservation law it is therefore possible to write the energy balance equation for a patch of ground as a sum of at least four energy flux densities (Thorn, 1975): + H + LE+G+...=0 (3) The Bowen ratio can be determined experimentally (Monteith and Unsworth, 1990) as the ratio of the vertical gradients of temperature (37") (or potential temperature) and humidity (Be): fi = y8T/8e (4) where y is the psyhrometric constant. The derivation of equation 4 assumes that the turbulent transport coefficients for sensible (H) and latent heat (LE) are equal, i.e. KH = KE (which is reasonable given that the dominant transport mechanism is by mechanical and convective turbulence). The fluxes can be determined by: H= LE = (1+/3" 1 ) (1+/3) (Rn-G) cppAT+XAq (7) if a direct flux measurement is available of H or LE (e.g. by eddy covariance), plus the relevant gradients, then: AC AC -= (8) Aq AT The Bowen ratio-energy balance approach can be applied in the roughness sublayer above vegetation as it only assumes similarity of exchange coefficient between scalars (Lenschow, 1995). Direct methods Eddy covariance: The eddy flux of any scalar can be written c = wPc (9) where Fc is the flux density of scalar c, w is the vertical wind speed, pc is the density (or concentration) of the scalar. The overbar represents the mean of the product over the sampling interval. Records of wind speed, temperature and concentration exhibit turbulent or irregular form; it is convenient to regard these variables as the sum of a mean and a fluctuating part. This process is known as Reynold's decomposition (Arya, 1988) and for wind speed and concentration can be written: w = w + w' (10) (5) Pc = Pc + Pc (11) (6) where the primes represent the fluctuation about the mean. Equation 9 can be re-written making use of this decomposition to produce The energy balance technique has been used extensively in the past for determination of sensible and latent heat fluxes over a range of crops and forests (Kim et al., 1989; Barr et al., 1994) and its accuracy is mainly dependent on the ability of instrumentation to measure small gradients of temperature and humidity, to measure the other components of the energy budget (including net radiation and soil heat fluxes) and in conditions which avoid local advection. It can be particularly difficult to use the technique when the net radiation values are small, for example, at sunrise and sunset. The energy balanceBowen ratio approach can be extended to determine the Ec = wp~c + wyc (12) which shows that the total vertical flux of any scalar is the sum of a mean vertical flux wpc and an eddy flux w'p'c. It will be noted that in the full expansion of equation 9, some terms involving the mean of a fluctuating component have been omitted from equation 12 because, by definition vv'and p'c = 0; terms involving the mean of the product of two fluctuating components, e.g. w'p'c will rarely be zero. One assumption normally made is that over a suitable interval of time there is no mass movement of the air in Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 P = H/LE surface-atmosphere exchange of trace gases making the assumption that the transfer characteristics for humidity, temperature and the trace gas species are identical. There are two formulations possible: If gradients are measured of temperature (AT) and humidity (Aq) are made simultaneously with gradients of trace gas species (AC): (2) where Rn is net all-wave radiative energy flux, H, sensible heat flux, LE, latent heat flux, and G is the soil heat flux density. These four fluxes usually account for about 90-95% of the total energy available (maximum photosynthetic fluxes could account for up to 3%; fluxes representing the horizontal transport of energy in advective situations could be greater, but if care is taken to use flat, extensive, homogeneous field sites, can usually be neglected). The determination of sensible and latent heat fluxes (H and LE) can be achieved by introducing the definition of the Bowen ratio (ft): 1135 1136 Moncrieff the vertical i.e. w = 0. With this proviso, the practical working equation for eddy covariance is obtained from equation 10, which is Fc = w'p'c + correction terms (13) (c) the entity in question must be averaged to some extent by the finite sampling volume of the sensor, for example, typically 20 cm between sonic transducers or between the source and detector of infra-red radiation in an openpath analyser, (d) the data acquisition hardware may not sample rapidly enough to capture high frequency eddies, (e) in the special case of using a closed-path analyser and ducting air, some of the turbulent structure will be lost, particularly the higher frequencies and again this will result in flux underestimation unless accounted for, (f) all the other usual micrometeorological requirements must also be met including stationarity, homogeneity of upwind surface and fetch (these issues have been well explored in many previous reviews such as Lenschow, 1995; Verma, 1990; Baldocchhi et al. 1988). Although this list can appear quite daunting, in practice, these corrections are well known and several papers describing eddy covariance systems give details, for example, Shuttleworth et al. (1988) and Moncrieff et al. (1997). Specific experiments have been reported which attempt to test rigorously the correction terms and the general feeling amongst micrometeorologists is that the corrections can be well quantified and can be relied upon (Suyker and Verma, 1993; Leuning and Judd, 1996). Fc=w'p'c + fjL(pJpa)LE + other correction terms (14) Instrumentation where n = mjmv, the ratio of the molecular masses of dry air to water vapour, pa = density of dry air, p,. = mean density of CO2 in air. This equation assumes that both water vapour and CO2 are brought to a common temperature and pressure within the optical bench. Leuning et al. (1982) provided experimental proof of the WPL corrections, Leuning and Moncrieff (1990) developed the arguments for a closed-path system and Leuning and King (1992) applied the equations to both open- and closed-path systems. The working equation shows that it is necessary to have instrumentation which can sample vertical wind speed and scalar concentration and be able to perform either real-time analysis in which means are subtracted from raw data to yield the fluctuating components then cross-products formed or have the facility to store all the raw data for later processing in the laboratory. Corrections The other corrections can be summarized as: (a) the sensor may not be capable of responding quickly enough to all of the flux-carrying eddies moving past it. In order to apply the eddy covariance equation, measurements of the fluctuating components of trace gas concentration and vertical wind speed are needed. Both vertical wind speed and concentration have to be measured with fast response sensors in order to catch the majority of the fluctuating components of the turbulent structures which are responsible for the fluxes. Sonic anemometers can measure the vertical velocity to an accuracy of about 0.5 cm s"1 and usually measure all three components of the turbulent wind, i.e. two horizontal and one vertical component. Modern sonic anemometers also measure the speed of sound and this can be used to infer air temperature and hence sensible heat flux (or more correctly and virtually identically, virtual heat flux, Schotanus et al., 1983). The concentration of carbon dioxide and water vapour in air can be measured by means of a rapidly responding infra-red gas analyser. Originally, only openpath analysers were available for fast response determinations of carbon dioxide and water vapour concentration (Ohtaki and Matsui, 1982). In these instruments, a source of infra-red radiation is separated from a detector by a path which is open to air to pass freely between. Technical Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 The assumption that w = 0 is problematic, but sufficient experimental evidence suggests that it can be corrected for in subsequent analysis. The issue is that sensible and latent heat fluxes cause variations in air density and most non-dispersive infra-red gas analysers measure the density of the trace gas species in air rather than the mixing ratio or mole fraction. The effect can be substantial for trace gases which have fluxes which are small in relation to their background concentration (Fowler and Duyzer, 1989). The need for a correction is also dictated by the type of measuring system in use, i.e. measuring CO2 by an open-path gas analyser may require corrections of 50-100% and opposite in sign simply due to simultaneous transfer of sensible heat in the middle of the day (Suyker and Verma, 1993). By contrast, in closed-path systems, temperature fluctuations can be effectively eliminated by passage down a sampling tube and the correction term for sensible heat flux is eliminated (although a correction for latent heat flux would remain unless the air was dried before entering the optical bench). A paper by Webb et al. (1980, hereafter referred to as WPL) showed how to calculate this effect and equation 14 is written for a typical closed-path IRGA, in this example, one measuring CO2 and H2O in the same instrument. The flux of CO2 would be (b) the sensor for w may be too far from the scalar sensor and flux loss can occur, Trace gas exchange The design of an eddy covariance system An eddy covariance system comprises not just the instrumentation, but also the software used to analyse the data and to correct for any flux loss imposed by the inevitable inability of the system to respond to all the flux-carrying eddies. Different design criteria need to be applied in systems employing open- or closed-path infra-red gas analysers (Leuning and Judd, 1996). Fortunately, a number of complete systems have been described in detail in the literature and it is not necessary to 're-invent the wheel' when it comes to putting together a system (Lloyd et al., 1984; Moncrieff et al., 1997; Grelle and Lindroth, 1996). These papers can be consulted to discover the necessary steps to be taken to correct for the inadequate sampling of any turbulent signal in an eddy covariance system. Figure 1 shows a schematic of a typical eddy covariance system in which air is ducted down a sampling tube from a point near the sensing volume of the sonic anemometer to a closed-path IRGA. Variations in flow rate are automatically compensated for by having an in-line pressure transducer providing relevant information to correction algorithms within the controlling software of the IRGA itself (Valentini et al., 1996). It is also possible to maintain flow rates at a fairly constant level by including a mass flow controller in the sampling line (Moncrieff et al., 1997). Evaluation of eddy covariance systems Filter Pressure Closed Path Analyzer LJ6262 Air pump O Flow meter PC Fig. 1. Schematic of a typical eddy covariance system. Air is drawn through a sample tube (with pre-filter) to the IRGAs optical bench by a pump; the sonic anemometer digitises the turbulence signals and also samples and digitizes the output from the IRGA, the process being controlled by software on a PC. heat transfer (G) and canopy heat storage (S)—is employed as a test in many eddy covariance flux studies (Baldocchi et al., 1988; Valentini et al., 1991; Greco and Baldocchi, 1996). The ability to close the surface energy balance, demonstrated by plotting net radiation against the sum of H, LE and G is one test for an eddy covariance system and closure over a sufficiently long period (days), can demonstrate that the system is working properly (Fig. 2). Typically, errors can account for variability on the order of 10-20%. Other tests are based on comparing the spectral and co-spectral response of individual instruments or the whole system against some well-tested Deciduous forest 1993-1994 250 -50 -50 The quality of eddy covariance data sets can be tested by examining the ability to close the surface energy balance and checking the spectral response of the turbulence measurements. Closure of the surface energy balance— the partitioning of net radiation (R^ into H, LE and soil 3D sonic 50 100 150 200 250 . ^ dally avaraga fWm" 2 ) Fig. 2. Typical test for an eddy covariance system showing the degree of energy balance closure. Integrated daily values of net radiation are plotted against the sum of sensible, latent and soil heat flux plus canopy heat storage. (Greco and Baldocchi, 1996). Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 improvements to more conventional closed-path analysers of the kind traditionally employed in plant physiological research has meant that eddy covariance systems based on this design have become more common in recent years. These instruments have optical benches with a smallvolume and through which air can be sampled at several litres per minute to achieve the necessary response time. Closed-path analysers are commercially available (few open-path IRGAs can be bought commercially), they can be easily calibrated with gas mixtures in situ and as air is brought to a common temperature before entering the optical bench, they require smaller corrections for air density effects (the WPL corrections). Perhaps just as crucially is the fact that the systems can be made weatherproof and can be left unattended for extended periods (Moncrieff et al., 1997; Goulden et al., 1996). The adoption of such technology means that micrometeorology is no longer regarded as a 'fair weather science'. 1137 1138 Moncrieff standard forms. Again, details of the tests which system builders have used appear in papers describing existing systems (Lloyd et al, 1984; Grelle and Lindroth, 1996; Moncrieff ef al, 1997). The implications of errors in eddy covariance systems and how they affect the long-term measured carbon balance of ecosystems are now also being addressed given that the technology is now mature enough to start monitoring over extended periods (Baldocchi et al, 1996). Particular tests for assessing the quality and reliability of surface-based flux measurements have been demonstrated both theoretically (Businger and Delaney, 1990; Foken and Wichura, 1996) and by reference to existing long-term datasets (Moncrieff et al, 1996). As an example of the long-term precision of eddy covariance systems to measure carbon sequestration over extended periods, the work done at Harvard forest in the US over several years, indicates that an uncertainty of + 5% might be expected in well-maintained and wellcharacterized eddy flux systems (Goulden et al, 1996). D144 1993 Whilst eddy covariance techniques overcome most of the problems involved with gradient measurements, it is applicable only to those gaseous species for which traditional fast response analysers are commercially available, effectively, CO2 and H2O. Thus, for example, measure- 1800 2400 600 1200 1800 2400 1200 1800 2400 2000 Time (hour*) Fig. 3. Daily trend of carbon dioxide flux (Fc), water vapour ( LE) and PAR over a mixed deciduous forest in Tennessee, USA, May 1993. Surface fluxes were obtained by eddy covariance. (Greco and Baldocchi, 1996) 1 i 22-05-93 24-05-93 Ll i U nn r- 20 - 1 23-05-93 21-05-93 10 - o A Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 Eddy accumulation 1200 350 Examples Figure 3 shows daily trends of the exchange of carbon dioxide, Fc photosynthetically active radiation, PAR, and latent heat, LE, between a mixed deciduous forest canopy (Tennessee, USA) and the atmosphere in May 1993 (Greco and Baldocchi, 1996). The magnitude of carbon dioxide uptake (negative values) increases with increasing PAR and achieves a peak value near — 2.0 mg CO2 m2 s"1 around midday. As the afternoon progressed, rates of carbon dioxide uptake slowly decrease until around 7.30 p.m., when respiration (positive values) exceeds net assimilation. Nocturnal rates of CO2 efflux are on the order 0.2 mg m2 s~\ The net daily carbon dioxide flux is -40.6 g CO2 m2 d"1 corresponding to 11.1 g C m" 2 d"1. For LE and PAR, significant rates of exchange commence at sunrise and peak values occur after midday. The peak values for LE and PAR are 340 W m~2 and 1700^mol m" 2 s"1, respectively. Figure 4 shows weekly variations of CO2 exchange rates, Fc, over a beech forest in Central Italy (Valentini et al, 1996). In this case, positive values are related to CO2 accumulation by the forest during the day and negative values of Fc represent night-time release of carbon dioxide due to forest respiration. The weekly averaged peak value for daytime CO2 flux density was about + 15 /xmol m~2 s~\ while the averaged night-time value was about -9 ^mol m~2 s~'. 600 -10 0 20 40 60 80 100 120 140 Time (hr) Fig. 4. Weekly variation in CO 2 exchange rate over a beech forest in Central Italy, May 1993. Positive values indicate net uptake by the forest. Trace gas exchange F x = jSoaH,(Cx+-C;c-) A Sonic anemometer Air Pump Solenoid valve B Absorbing traps (15) where |3O is a semi-empirical coefficient (with a typical value of about 0.56), ow the standard deviation of vertical wind speed and Cx and C~, the concentration of the gas in the up- and down-reservoirs, respectively. The basic principle of the technique remains that the air should be sampled according to the sign of the vertical wind speed, but the sampling is done simply on the condition of the vertical wind, i.e. either up or down. Relaxed eddy accumulation (REA) is now popular and a number of designs have appeared in the literature for compounds such as CO 2 , CH 4 , NO 2 , non-methane hydrocarbons, and pesticides (Beverland et al., 1996a, b; Majewski et al., 1993). The WPL corrections may be required with this technique according to the degree of pre-processing of the air streams (Pattey et al., 1992). The relative errors in this technique are similar to those in eddy covariance as has been shown by a number of studies which use both techniques simultaneously (Fowler et al., 1995). Figure 5a shows a basic REA scheme with one solenoid valve, two Teflon bags for collection of the gas and a sonic anemometer. Not all chemical species can be measured in exactly the same way, however. For example, the determination of VOCs requires particular attention, such as the use of a pump at the top of the line and bags, whose internal surfaces have to be previously well cleaned in order to avoid any contamination. A second scheme is based on the use of collecting samples on chemical absorbers Solenoid valves Air pump Fig. 5. (a) A schematic of a basic configuration for an eddy accumulation system with one solenoid valve, two teflon bags for sample collection and a sonic anemometer, (b) As above but with absorbing tubes rather than sampling bags. instead of in bags (Fig. 5b). The advantage of this configuration is it avoids the problem of contamination since air is drawn through the two traps which collect air according to the sign of vertical wind speed (+ up-draft, - down-draft). When air is sampled through one trap, the other solenoid valve is open to the external air. Improvements and innovations are regularly being suggested and these ensure that relaxed eddy accumulation will find increasing favour in biosphere-atmosphere experiments (Oncley et al., 1993). Figure 6 shows an example of flux determination with a relaxed eddy accumulation system. The graph shows the diurnal trend of a-pinene flux emitted by a mixed Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 ments of Volatile Organic Compound's (VOC's) are not possible by eddy covariance. Developments with tunable diode laser systems promise some widening of the scope for eddy covariance, given that fluxes of methane and nitrous oxide can now be measured successfully (Edwards et al., 1994; Fowler et al., 1995), but its application to other trace gases is still restricted. A technique related to eddy covariance is rapidly gaining acceptance as it swaps the difficulty of making fast analysis of the chemical species for the much more practical high-precision of slower responding chemical analysers, The technique is known as eddy accumulation or conditional sampling. In its original form (Desjardins, 1977) air was sampled into two separate reservoirs according to the magnitude and direction of the vertical wind. Although the technique was used it was never a great success because of the difficulty of collecting rapid samples at a rate proportional to the magnitude of the vertical velocity (Speer et al, 1985). A practical solution has been the use of a 'relaxed' form of the initial eddy accumulation theory in which an empirical coefficient is introduced which effectively eliminates the need to sample at variable rates. This form takes the name of relaxed eddy accumulation (Businger and Oncley, 1990) and can be expressed as follows for the flux Fx of a chemical species x: 1139 1140 Moncrieff a - pincne density flux over a pine - oak forest 1.0 + 1 1 1 i 0.8 - - 0.6 0.4 - \ 0.2 0 0 H» h—m i i 10 15 20 hours Fig. 6. Trend of a-pinene over a mixed pine-oak forest near Castelporziano, Italy using the relaxed eddy accumulation system shown in Fig. 5b. Conclusions Fetch: the concept of a 'flux footprint' Fluxes measured by micrometeorological sensors are effectively the integration of fluxes from a variety of sources and sinks in the landscape for a distance of several hundred metres upwind from the measuring point. The height at which the measurements are chosen to be made must be determined both by a consideration of the frequency response of the instrumentation and also the 'fetch' or extent of the upwind area which is uniform. Eddies become progressively larger with height up to the depth of the planetary boundary layer, typically 1 km by day, and this means that instrumentation with a slower response can be used successfully at heights well above the vegetation. As the surface is approached, however, the spectrum of turbulence includes a greater proportion of smaller eddies which will be actively exchanging mass and momentum between the surface and the atmosphere. The instrumentation then must be capable of sampling the high frequency eddies, typically up to 10 Hz. In principle, whilst it may be less difficult to measure surface fluxes using the eddy covariance technique with height, in practice, the difficulty lies in ensuring that the boundary layer that is being measured is the one representative of the vegetation upwind. If the instruments are placed too high above the surface it is possible that they could extend out of the boundary layer representative of the upwind vegetation and instead be measuring some component of fluxes from a different type of vegetation some greater distance upwind. A convenient rule-of-thumb suggests a Continued developments in both theory and instrumentation suggest that making flux measurements by micrometeorological methods will continue to play a central role in studying land-atmosphere exchange of radiativelyactive gases. Traditional methods based on establishing exchange coefficients will continue to be used but, increasingly, it will be possible to make routine, long-term flux measurements by eddy covariance and eddy accumulation methods. This is essential to ensure the questions of global change biology are addressed and that the basic data are available to parameterize and test simulation models. References Arya SP. 1988. Introduction to micrometeorology. San Diego: Academic Press. Baldocchi DD, Hicks BB, Meyers TP. 1988. Measuring biosphere-atmosphere exchanges of biologically related gases with micrometeorological methods. Ecology 69, 1331-40. Baldocchi DD, Valentin) R, Running S, Oecfael W, Dahlman R. 1996. Strategies for measuring and modelling carbon dioxide and water vapour fluxes over terrestrial ecosystems. Global Change Biology 2, 159-68. Ban- AG, King KM, GUlespie TJ, den Hartog G, Neumann HH. 1994. A comparison of Bowen ratio and eddy correlation sensible and latent heat flux measurements above deciduous forest. Boundary-Layer Meteorology 71, 21-41. Beverland IJ, O'Neill DH, Scott SL, Moncrieff JB. 1996a. Design, construction and operation of flux measurement systems using the conditional sampling technique. Atmospheric Environment 30, 3209-20. Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 pine-oak forest of the Mediterranean region (Castelporziano, Italy). The measurements have been performed with the system described in Fig. 5b with absorption traps built with a special formulation of carbon. The trend follows a typical daily course with a slight deposition during the night and a midday reduction, probably related to physiological constraints. fetch:height ratio of about 100:1; thus a fetch of 500 m would allow instruments to be placed up to a height of about 5 m above the surface. The fetch:height ratio will depend on atmospheric stability and surface roughness in so far as they influence the degree of mixing of internal boundary layers as they are advected over different types of surface (Gash, 1986). The concept of'fetch' is one which continues to exercise micrometeorologists and plant physiologists seeking to compare leaf or chamber (essentially point measurements) with the spatially averaged fluxes measured by towerbased micrometeorological techniques. One promising approach is to define a 'footprint' or source region which is a measure of the relative importance of sources upwind which contribute to the measured flux. This area can be regarded as contributing most of the flux measured and its areal extent and position can be calculated from a knowledge of surface roughness, atmospheric stability, wind speed and direction. Schemes such as those proposed by Schuepp et al. (1990) and Schmid (1994) permit the calculation of the flux footprint under a range of surface and atmospheric conditions. Trace gas exchange Leuning R, Denmead OT, Lang ARG, Ohtaki E. 1982. Effects of heat and water vapour transport on eddy covariance measurement of CO 2 fluxes. Boundary-Layer Meteorology 23, 209-22. Leuning RL, Judd M. 1996. The relative merits of open- and closed-path analysers for measurement of eddy fluxes. Global Change Biology 2, 241-54. Leuning RL, King KM. 1992. Comparison of eddy covariance measurements of CO 2 fluxes by open- and closed-path CO 2 analysers. Boundary-Layer Meteorology 59, 297-311. Leuning RL, Moncrieff JB. 1990. Eddy covariance CO 2 flux measurements using open- and closed-path CO 2 analysers: corrections for analyser water vapour sensitivity and damping of fluctuations in air sampling tubes. Boundary-Layer Meteorology 53, 63-76. Lloyd CR, Shuttleworth WJ, Gash JHC, Turner M. 1984. A microprocessor system for eddy correlation. Agricultural and Forest Meteorology 33, 67-80. Majewski MS, Desjardins RL, Rochette P, Pattey E, Seiber JN, Glotfelty DE. 1993. A field comparison of an eddy accumulation and an aerodynamic gradient system for measuring pesticide volatilization fluxes. Journal of Environmental Science and Technology 27, 121-8. Moncrieff JB, Malhi Y, Leuning RL. 1996. The propagation of errors in long-term measurements of land-atmosphere fluxes of carbon and water. Global Change Biology 2, 231-54. Moncrieff JB, Massbeder JM, DeBruin H, Elbers J, Friborg T, Huesunkveld B, Kabat P, Scott SL, Soegaard H, Verhoef A. 1997. A system to measure surface fluxes of momentum, sensible heat, water vapour and carbon dioxide. Journal of Hydrology (in press). Monteith JL, Unsworth MH. 1990. Principles of environmental physics, 2nd edn. London: Arnold. Ohtaki E, Matsui T. 1982. Infra-red device for simultaneous measurement of fluctuations of atmospheric carbon dioxide and water vapour. Boundary-Layer Meteorology 24, 109-19. Oncley SP, Delaney AC, Horst TW. 1993. Verification of flux measurements using relaxed eddy accumulation. Atmospheric Environment 27, 2417-26. Pattey E, Desjardins RL, Boudreau F, Rochette P. 1992. Impact of density fluctuations on flux measurements of trace gases: implications for the relaxed eddy accumulation technique. Boundary-Layer Meteorology 59, 195-203. Raupach MR. 1988. Canopy transport processes. In: Steffen WL, Denmead OT, eds. Flow and transport in the natural environment: advances and applications. Berlin: SpringerVerlag, 95-127. Raupach MR, Thorn AS. 1981. Turbulence in and above plant canopies. Annual Reviews in Fluid Mechanics 13, 97-129. Schimel SD. 1995. Terrestrial ecosystems and the carbon cycle. Global Change Biology 1, 77-91. Schmid HP. 1994 Source areas for scalars and scalar fluxes. Boundary-Layer Meteorology 67, 293-318. Schotanus P, Nieuwstadt FTM, de Bruin HAR. 1983. Temperature measurements with a sonic anemometer and its application to heat and moisture fluxes. Boundary-Layer Meteorology 26, 81-93. Schuepp PH, Leclerc MY, Macpherson JI, Desjardins RL. 1990. Footprint prediction of scalar fluxes from analytical solutions of the diffusion equation. Boundary-Layer Meteorology 50, 355-76. Shuttleworth WJ, Gash JHC, Lloyd CR, McNeil DD, Moore CJ, Wallace JS. 1988. An integrated micrometeorological system for evaporation measurement. Agricultural and Forest Meteorology 43, 295-317. Speer ER, Peterson KA, Ellestad TG, Durham JL. 1985. Test Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 Beverland IJ, Scott SL, O'Neill DH, Moncrieff JB, Hargreaves K. 19966. Simple battery powered device for flux measurements by conditional sampling. Atmospheric Environment 31, 277-81. Businger JA, Delaney AC. 1990. Chemical sensor resolution required for measuring surface fluxes by three common micrometeorological techniques. Journal of Atmospheric Chemistry 10, 399^110. Businger JA, Ondey SP. 1990. Flux measurement with conditional sampling. Journal of Atmospheric and Ocean Technology 7, 349-52. Desjardins RL. 1977. Description and evaluation of a sensible heat flux detector. Boundary-Layer Meteorology 11, 147-54. Edwards GC, Neumann HH, den Hartog G, Thurtell GW, Kidd G. 1994. Eddy correlation measurements of methane fluxes using a tunable diode laser at the Kinosheo Lake tower site during the Northern Wetlands Study (NOWES). Journal of Geophysical Research 99, 1511-17. Foken Th, Wichura B. 1996. Tools for quality assessment of surface-based flux measurements. Agricultural and Forest Meteorology 78, 83-105. Fowler D, Duyzer JH. 1989. Micrometeorological techniques for the measurement of trace gas exchange. In: Andrae MO, Schimel DS, eds. Exchange of trace gases between terrestrial ecosystems and the atmosphere. New York: John Wiley & Sons, 189-207. Fowler D, Hargreaves KJ, Skiba U, Milne R, Zahniser MS, Moncrieff JB, Beverland IJ, Gallagher MW. 1995. Measurement of CH 4 and N 2 O fluxes at the landscape scale using micrometeorological methods. Philosophical Transactions of the Royal Society of London, Series A 350, 339-56. Fowler D, Unswortta MH. 1979. Turbulent transfer of sulphur dioxide to a wheat crop. Quarterly Journal of the Royal Meteorological Society 105, 767-84. Gash JHC. 1986. A note on estimating the effect of limited fetch on micrometeorological evaporation measurements. Boundary-Layer Meteorology 35, 409-14. Goulden ML, Munger, JW, Fan S-M, Daube BC, Wofsy SC. 1996. Measurements of carbon sequestration by long-term eddy covariance: methods and a critical evaluation of accuracy. Global Change Biology 2, 169-82. Greco S, Baldocchi DD. 1996. Seasonal variations of CO 2 and water vapour exchange rates over a temperate deciduous forest. Global Change Biology 2, 183-97. Grelle A, Lindroth A. 1996. Eddy-correlation system for longterm monitoring of fluxes of heat, water vapour and CO 2 . Global Change Biology 2, 297-307. Horst TW, Weil JC. 1994. How far is far enough? The fetch requirements for micrometeorological measurements of surface fluxes. Journal of Atmospheric and Oceanic Technology 11, 1018-25. Kaimal JC, Finnegan JJ. 1994. Atmospheric boundary layer flows. Their structure and measurement. Oxford University Press. Kim J, Verma SB, Rosenberg NJ. 1989. Energy balance and water use in cereal crops. Agricultural and Forest Meteorology 48, 135-47. Lenschow DH. 1995. Micrometeorological techniques for measuring biosphere-atmosphere trace gas exchange. In: Mattoon PA, Harris RC, eds. Biogenic trace gases: measuring emissions from soil and water. Oxford: Blackwell Science, 126-63. Lenschow DH, Mann J, Christens L. 1994. How long is long enough when measuring fluxes and other turbulence statistics? Journal of Atmospheric and Oceanic Technology 11, 661-73. 1141 1142 Moncrieff of a prototype eddy accumulator for measuring atmospheric vertical fluxes of water vapour and paniculate sulfate. Journal of Geophysical Research 90, 2119-22. Sutton MA, Fowler, D, Moncrieff JB. 1993. The exchange of atmospheric ammonia with vegetated surfaces. I Unfertilized vegetation. Quarterly Journal of the Royal Meteorological Society 119, 1023-45. Suyker AE, Verma SB. 1993. Eddy correlation measurements of CO 2 flux using a closed-path sensor: theory and field tests against an open-path sensor. Boundary-Layer Meteorology 64, 391-407. Thorn AS. 1975. Momentum, mass and heat exchange. In: Monteith JL, ed. Vegetation and the atmosphere. Chichester: Academic Press, 57-109. Valentioi R, Scarascia Mugnozza GE, De Angelis P, Bimbi R. 1991. An experimental test of the eddy correlation technique over a Mediterranean macchia canopy. Plant, Cell and Environment 14, 987-94. VaJentini R, De Angelis P, Matteucci G, Monaco R, Dore S, Scarascia Mugnozza GE. 1996. Seasonal net carbon dioxide exchange of a beech forest with the atmosphere. Global Change Biology 2, 199-207. Verma SB. 1990. Micrometeorological methods for measuring surface fluxes. Remote Sensing Reviews 5, 99-115. Webb EK, Peannan GI, Leuning RL. 1980. Correction of flux measurements for density effects due to heat and water vapour transfer. Quarterly Journal of the Royal Meteorological Society 106, 85-100. Downloaded from http://jxb.oxfordjournals.org/ at EC JRC Ispra site on March 7, 2013 View publication stats