Soil and Environmental Analysis: Modern Instrumental Techniques - Chapter 11 ppsx

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, and

to investigate some issue of soil–vegetation–atmosphere exchange This chapter discusses the methods that can be used at the scale of the agricultural landscape The methods belong to the general field of micrometeorology in that they have time and space scales that are on the order of tens of minutes and a few km 2 respectively The space scale is influenced both by the length scales of atmospheric turbulence, which are

a result of mechanical (surface friction) effects, and by thermal effects which influence atmospheric stability The reporting time scale for surface fluxes of about 30 min is related to the need to make observations over a suitably long period, so that the majority of the spectra of flux-carrying eddies are sampled, yet not so long that natural diurnal variabilities in scalar concentrations, or forcing functions such as solar radiation, are included One important difference between micrometeorological methods and other techniques to measure surface fluxes is that they are nondestructive

in that they only sample the air as it advects past the sensor; they also are noncontact in that we merely sample passively as the air passes our instrumentation—they cannot alter the microclimate as chamber methods,

for instance, could do (see Chap 10) Traditionally, micrometeorologists have sought to establish their measuring systems on landscapes that are

as flat as possible and with homogeneous surface cover as far as possible upwind The earliest micrometeorological experiments were made over surfaces such as extensive wheat fields or prairie grass Measurements were made generally in good weather, partly to protect the type of instrumenta- tion then available In the past decade, instrumentation has developed to the extent that routine flux measurements are possible in all weather and for long periods of time There has been a proliferation of routine flux stations across the globe (although irregularly distributed in space), and this has inevitably led to many of the stations being in nonideal terrain, e.g., on the gentle slopes of hills The challenge for micrometeorology in the next decade

or so is to develop the theoretical issues that such an expansion of sites brings.

Micrometeorological methods can be used to scale up to observations made by aircraft or interpolated from remote sensing platforms They can also be used to check observations made on much smaller scales, e.g., chambers on leaves or soil To that extent they are an integral part of the observation strategy that is being used to solve many of the current environmental problems facing us It is the scaling issues that bring micrometeorologists and other observational and modeling scientists together.

II THEORY AND SCALES OF OBSERVATION

A Where the Observations Are Made

All micrometeorological measurements are made within the atmospheric boundary layer (ABL, Fig 1), defined 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.’’ The depth of the ABL depends on the degree of mechanical (caused by surface friction) and buoyant mixing (thermals rising from the warmed surface), and its depth can also be dictated by synoptic scale motions such as anticyclonic subsidence The ABL is sometimes also called a convective boundary layer (CBL) when it is several kilometers deep as a result of the development

of thermals rising during the day At night, the ABL may be perhaps only a few tens of meters deep The top of the ABL represents a fairly sharp boundary between the turbulent, chaotic motions of the ABL and the smoother, streamlined flow of the free atmosphere above The rate of change in vertical profiles of temperature, water vapor, and carbon dioxide

is highest near the active surface, in a region typically about one-tenth the

depth of the ABL that is called the surface boundary layer (SBL) It is in this region that most conventional micrometeorological measurements are made The SBL can further be divided into two sublayers: inertial and roughness (Raupach and Thom, 1981) In the inertial sublayer, wind profiles in neutral stability conditions are logarithmic with height, and well-established scaling schemes apply (Kaimal and Finnegan, 1994) Fluxes are considered constant with height (or at least within 10% of their surface values) Close to surface vegetation lies an interfacial layer in which turbulence is enhanced over

that in the inertial sublayer above by wake turbulence or thermal effects The depth of this so-called roughness sublayer has been variously estimated

to be three times the height of the vegetation (Kaimal and Finnegan, 1994)

or sometimes three times the spacing between the vegetation elements (Raupach and Legg, 1984) The implication of the enhanced diffusivities in the roughness layer is that micrometeorological methods that rely on establishing eddy diffusivities are difficult to apply in this layer By going beyond the roughness sublayer, however, concentration gradients can become small and difficult to measure over rough vegetation, and this poses further problems.

The top of the surface layer is not physically as well defined as the top

of the ABL Although most tower-based flux measurements are made within the surface boundary layer, the evolving structure of the ABL over a day does present opportunities for other measurements using aircraft, tether- sondes, and ABL budget methods (Raupach et al., 1992).

B 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 meters upwind from the measuring point The height at which the measurements are chosen to be made must be determined both by consideration of the frequency response of the instrumentation and also by the ‘‘fetch’’ or extent of the upwind area from which the signal comes 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 larger proportion of smaller eddies that actively exchange mass and momentum between the surface and the atmosphere Instrumentation used in the eddy covariance method must therefore be capable of sampling high frequency eddies, typically up to 10 Hz In principle one could use an eddy covariance system (see Sec III.C) well above the canopy in order to avoid the problem of frequency response of analyzers However, as we move up in height the area of flux integration becomes larger and the requirements of surface homogeneity become more and more stringent In fact if the instruments are placed too high above the surface it is possible that they could extend above the boundary layer representative of the nearby vegetation and be measuring some component of fluxes from a different type of vegetation further upwind A convenient rule of thumb suggests a 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 depends on atmospheric stability and surface roughness insofar as they influence the degree of mixing of internal boundary layers as they are advected over different types of surface (Mulhearn, 1977; Gash, 1986; Grelle and Lindroth, 1996) The footprint

or source region defines the relative importance of sources upwind that 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, and wind speed and direction (e.g., Schuepp et al., 1990, Schmid and Oke, 1990) The approach is based on the same theory as underlies dispersion modeling of pollutants using the Gaussian plume approach Such models can be used to calculate the relative contribution to the vertical flux at any measurement height coming from a point upwind Figure 2 is an example

of the distribution of the relative contribution to the vertical flux as a

flux contributed by sources at various distances upwind from the measuring point (The simulation is available at http://www.ierm.ed.ac.uk/jbm/java/jflux/jf2.htm.)

function of height of measurement The data are normalized so that the flux at the distance of maximum source contribution appears as a peak

in this representation In this simulation, the peak source distribution is within about 50 m of the measurement point The contribution from sources upwind decreases exponentially with distance from the tower Calculation of the cumulative fraction shows that even at a distance of about 1500 m from the tower, only about 95% of the measured flux has been accounted for by sources within this distance or footprint Both the peak and the cumulative fraction depend on measurement height and atmospheric stability As the height of measurement increases, the peak of the flux footprint becomes more and more distant from the point of measurement Similarly, the peak contribution moves closer to the point of measurement as the atmosphere becomes more unstable With increasing stability, the peak contribution moves further from the point

of measurement.

Many agricultural landscapes are characterized by relatively scale heterogeneity (small fields of a few hundred meters on a side) and thus any tower-based flux measuring system will ‘‘see’’ fluxes coming from different fields within the footprint The flux footprint concept permits the integration of fluxes from such a landscape by permitting the calcula- tion of the relative source strengths in the area upwind of the flux tower Soegaard et al (2003) examined the CO 2 fluxes over several very different agricultural surfaces (winter wheat, winter barley, spring barley, maize, and grass) in conjunction with a footprint modeling exercise Good agreement was found between the surface flux measured on a tower where

small-it could be expected to integrate across the different land-use types and the modeled fluxes (based on 3-D footprint and biophysical models) Such

an approach is likely to be of increasing value in the real world where homogeneous extensive landscapes for micrometeorological research are really quite rare.

C Measurements of Net Ecosystem Exchange

Fluxes of gases measured by micrometeorological methods above the canopy are the net fluxes from the whole system If the gases of interest are carbon dioxide and water vapor then the net fluxes are measures of the net ecosystem exchange of carbon (NEE) or total evaporation (ET), respec- tively, from the canopy, its elements and the ground surface (Fig 3) The upper system boundary (USB) marks the level through which the vegetation exchanges carbon and water with the atmosphere Direct micrometeoro- logical methods such as eddy covariance operate at this level Below the

USB, A l and R l are the net assimilatory and respiratory exchanges by the leaves The subscripts w, s, and r refer to respiratory fluxes from wood, soil, and roots, respectively Instrumentation placed above the upper system boundary (USB) measures the net exchange of material passing through that arbitrary level and of itself cannot distinguish the pathways by which that flux arrived at the sensor Thus, the NEE is the measured flux (F c ) above the USB, plus the component that represents storage of carbon between the point of observation and the ground, i.e.,

Fc+
S (It is assumed that there are no advective fluxes in this representation; this assumption will be examined later.) Profiles of carbon dioxide concentration within the canopy, and up to the height at which the flux measurements are made, are used to measure changes in storage of carbon.

NEE is the net sum of a number of component processes that take place within the stand These include the gains of carbon in photosynthesis

by the foliage of the trees, understorey, and mosses, and the losses from respiration by the above-ground foliage and wood as well as the below- ground roots, mycorrhizas, and heterotrophic microorganisms (the so-called

‘‘soil respiration’’) Of these the major components are the gains by photosynthesis and the losses through soil respiration.

III PRINCIPLES

The transport of gases, heat, and pollutants in the atmosphere is produced

by the eddying motion of the atmosphere as air parcels are displaced from one level to another Micrometeorological methods used to quantify this turbulent exchange can either sample the air as it flows past a sampling point for its vertical windspeed and direction and its gas concentration directly (the eddy covariance or eddy accumulation methods), or they can be based on quantifying the rate of diffusion down concentration gradients (the

aerodynamic and Bowen ratio methods) The direct method of eddy covariance involves sampling at one height only but with relatively sophisticated sensors and logging equipment The methods based on measuring gradients require measurements at two or more heights but use simpler sensors The disadvantage of the gradient techniques is, however, that a number of empirical functions are required to account for thermal stratification of the atmosphere; additionally, the gradients in atmospheric properties become very small above vegetation canopies, and the aerodynamic technique in particular cannot be used inside plant canopies All three 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 of the vegetation, i.e., these methods cannot be used on isolated plots or small fields If these conditions are met, 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.

The question of which method to use depends not only on the available resources but also crucially on the surface type over which the measurements are to be made (Moncrieff et al., 2000) For example, over very rough surfaces in an aerodynamic sense such as forests, gradients

of scalars are small, and their measurement places extreme emphasis on the precision of sensors Under these conditions, gradient techniques are problematic On the other hand, as turbulence is enhanced over forests, the eddy covariance technique is made easier as the size of the fluctuations in vertical windspeed and other atmospheric properties is increased Measure- ments over smooth surfaces such as water or ice are difficult using any of the techniques mentioned Increasingly, eddy covariance-based systems are being integrated into global observing networks both to monitor carbon and water exchange on the global scale and to validate remotely sensed products such as fraction photosynthetically active radiation (FPAR) and normalized difference vegetation index (NDVI) both of which can be used to estimate

leaf area index, which in turn can be used to model NEE (Running et al., 1998).

A Aerodynamic Gradient Method

The vertical exchange of atmospheric entities such as momentum, temperature, water vapor, and CO 2 by turbulent transport is driven by and is proportional to their vertical concentration gradients We can describe the transport process in a flux-gradient form which defines the constant of proportionality K known as an eddy diffusivity In generic form, the flux density (F x , the amount of that entity transported vertically through unit area in unit time) of any scalar (X) is

Fx¼
Kx

dX dz The eddy diffusivities for scalars such as temperature, water vapor, and CO 2 have been the subject of much experimentation over several decades, as their relative magnitudes depend on both surface and atmospheric features In practice, the diffusivities are related to the eddy diffusivity for momentum, which can be established from profiles of wind speed Wind profiles are used to calculate the friction velocity, a measure of the degree of atmospheric mixing (Thom, 1975) and from which we find

within the canopy as the variation in sources and sinks for heat, water vapor, and carbon dioxide invalidates the underlying assumptions in the method (Raupach and Legg, 1984).

B Energy Balance/Bowen Ratio Method

The energy balance at the surface is

Rn
G
S ¼ H þ LE

where R n is net radiation absorbed by the vegetation, G is soil heat flux, and

S is heat stored in the vegetation The ratio of sensible (H) to latent heat (LE ) flux is known as the Bowen ratio (), and by writing the fluxes in their flux-gradient form, an equation can be found that permits either flux to be found from measurements of the gradient of temperature (T ) and humidity (e), irrespective of atmospheric stability:

This method can also be used to measure fluxes of other gases or pollutants

by writing a more generalized form of the flux-gradient equation by combining H and LE as before to yield

Rn
G
S ¼ Kcp

dT e

dz where K is an eddy diffusivity assumed equal for all entities (other than momentum) and T e is the equivalent temperature (T þ e/) (Monteith and

Unsworth, 1990) If all pollutants and gases share this value of K, then the flux density for any scalar, as determined by the energy balance method, can be written as

Fx¼ ðRn
G
SÞ dx

dT e

The gradient for any gas x can be found by plotting the concentration of x

in air against T e at the same height for a number of measurement levels The available energy (R n
G
S ) can be found using net radiometers, and soil heat flux plates, and from profiles of temperature within the canopy The technique is reliable in most conditions, but when the available energy becomes small, e.g., at night, or the gradients are small, as over rough vegetation, then large errors can occur Also the method involves measuring G and S, which poses problems In low-windspeed conditions, where the aerodynamic method fails because of the stalling of anemo- meters, the Bowen ratio method works well The Bowen ratio method can also be used in the roughness sublayer because it only requires the assumption of equality between the eddy diffusivities for sensible heat and water vapor, i.e., there is no need for stability factors invoking similarity with K M as are needed with the aerodynamic method (Lenschow, 1995).

C Eddy Covariance

The eddy covariance technique measures the flux of a scalar (heat, mass) or momentum at a point centered on instruments placed at some height above the surface The principal instruments are (a) a fast gas analyzer, with

a response time < 0.1 s, and (b) a sonic anemometer capable of measuring the three-dimensional components of the wind and with a similar time resolution to the gas analyzer For these measurements to be identical to the flux at the underlying surface, the instruments must be located in the internal boundary layer where the flux is constant with height Figure 4

shows a typical time series of the turbulent signals of vertical wind speed and

CO 2 measured over a forest We can split any of the turbulent signals into

a mean and d fluctuating part, in the form of a Reynold’s decomposition Thus, for vertical wind speed, the instantaneous value (w) is a function

of both the long-term mean value (w) and the instantaneous difference (w0) between that value and w, i.e., w ¼ w þ w0 We can write the signal for the scalar density in the same format,  c¼cþ0

c The vertical transfer

of the scalar is simply the product of the two fluctuating terms w0 and 0

c

averaged over a suitable interval of time (usually 10–30 min).

Fc¼w00

cþcorrections where F c is the eddy flux of a scalar such as carbon dioxide There are many papers that discuss the ‘‘correction’’ terms in detail, so we will not be concerned with such detail here Good detailed reviews of eddy covariance appear in Aubinet et al (2000) and Baldocchi (2003).

Eddy covariance instrumentation is usually placed at some distance above the source of trace gas so the storage of that gas below the instruments must be considered This is particularly important during stable atmospheric conditions, usually at night, when turbulent mixing is light or nonexistent; gas effluxing from the soil below a canopy, for instance, may be stored in the canopy below the instruments The contribution to the total flux from storage is

instantaneous concentration of a scalar, c and the instantaneous vertical wind speed, w are shown Mean values of these quantities over a time period, typically 10–30 min, are indicated by an overbar Fluctuations from the mean are indicated by a prime.

that can sample vertical wind speeds and scalar concentrations and be able either to perform real-time analysis, in which means are subtracted from raw data to yield the fluctuating components from which cross products are formed, or to store all the raw data for later processing in the laboratory.

Figure 5 shows the components of a typical eddy covariance system A sonic anemometer above the canopy measures the turbulent fluxes of horizontal and vertical wind speeds Air is sucked down an inlet tube near the sonic head to a fast-responding infrared gas analyzer at the base of the tower The expanded schematic shows the gas path within the gas analyzer A mass flow controller and pressure transducer can be used to maintain a constant rate

of flow down the sample tube (and hence constant lag of gas sample between the sonic head and the optical bench of the IRGA) Gas concentrations in the sample cell are measured relative to a reference cell in which the air has been dried and scrubbed of carbon dioxide and water vapour.

The size of the storage correction can be small or negligible during the day, but in tall vegetation such as forests, it can be substantial at night A correction term for nonzero mean velocity or convergence arises at even seemingly ideal field sites for micrometeorology Atmospheric subsidence is common in highly convective conditions and in synoptic-scale subsidence; local circulations such as lake breezes (Sun et al., 1998) and katabatic drainage even on slopes as little as 1 : 1000 can induce a nonzero mean vertical velocity, as drainage flow on a slope is compensated for by a descending motion According to Lee (1998), a mean vertical velocity of

3 cm s
1 could be generated on a slope of only 1 The implications for long-term flux measurement over a forest site of a trace gas such as CO 2 are important Lee (1998) showed that the mass flow mechanism may contribute

as much as 12 mmol m
2 s
1 with a mean vertical velocity of only 0.5 cm s
1

He also gives the example of a typical flux site with a slope of 1, which induced a mean vertical velocity at night of 1 cm s
1 , sufficient to induce a mass flux that was twice the size of the measured eddy flux As Lindroth

et al (1998) point out, for their long-term flux site in an old-growth spruce plantation in Sweden, a 1 mmol m
2 s
1 bias averaged over 12 h per day for

365 days would be equivalent to 200 g C m
2

y
1 and large enough to change the sign of the net carbon flux over this period Since the size of the correction term is proportional to height, the problem is not so large over short canopies such as cereals.

The instruments used in the eddy covariance method are now sufficiently robust, reliable, and waterproof that they can be used to measure surface fluxes for extended periods of time Figure 6 shows CO 2 fluxes for five years from 1997 at a forest in the Highlands of Scotland Photosynthesis dominates over respiration (the heavy black line represents

0 flux) during the summer months, but even in the winter there are periods

Figure 5 A schematic of a typical eddy covariance system.

during the day when C is being fixed by the canopy Data such as shown in Fig 6 can be used to estimate the total C sequestered by forests over long periods of time (in this instance, typical rates of sequestration averaged over the five years are about 6.5 t C ha
1

yr
1 ) The accuracy of such long-term results is difficult to judge, at least for trace gases, since there is no other check against which to compare the measurements One accepted method is

to check whether the other instruments usually deployed as part of the same micrometeorological experiment achieve energy balance, i.e., whether the sum of available energy [net radiation (R n ) minus heat taken up by the soil

(G)] balances the losses of energy through sensible (H) and latent heating (LE) of the air) Figure 7 shows results from the same forest as shown in

Fig 6 and for the same period, but for energy balance closure The data show that about 95% of the energy was accounted for by the instruments and analysis used to obtain H and LE (also by eddy covariance), thus suggesting that the CO 2 flux data should have been calculated to a similar level of accuracy (Moncrieff et al., 1996).

D Relaxed Eddy Accumulation

Relaxed eddy accumulation (REA), also known as conditional sampling, is

a conceptually simple micrometeorological technique in which air is sampled into one of two sampling reservoirs (or sampling lines for on-line analysis), according to whether an up- or a down-draught of air is measured simultaneously by a sonic anemometer After a suitable interval of time, say 30–60 min, the net vertical flux of the trace gas species of interest is proportional to the difference in gas concentration between the sampling reservoirs The method is attractive, as it can be used for trace gases or pollutants for which no suitable fast-response sensor is available and hence the alternative technique of eddy covariance is unsuitable (e.g., Majewski

et al., 1993) The requirement for a fast sensor to measure vertical wind speed remains, but there is considerable relaxation in the speed requirement for the chemical analyzer A further advantage is that by accumulating gas into reservoirs, the difference in concentration between the samples is enhanced, and it is then possible to use high-precision gas analyses in the laboratory to determine the differences (Businger and Delaney, 1990).

Early studies using eddy accumulation sampled air at a rate proportional to the magnitude of the vertical wind speed, but this was technically difficult to achieve with the required accuracy (Desjardins, 1977) Hicks and McMillen (1984) suggested, almost as an aside, that the simpler method of sampling air at a constant rate into either bag might work, and with fewer practical difficulties One bag will then contain air collected in updraughts and the concentration of CO 2 , say, will be cþ

; the downdraught bag will have a CO 2 concentration of c

The idea was taken up by Businger and Oncley (1990) who wrote the flux (F c ) for a gas with concen- tration c, as

Fc¼wðcþ
c
Þ

where  is an empirical coefficient usually determined by experiment, s w is the standard deviation of the vertical windspeed, and (cþ
c
) is the gas concentration difference between the two sampling bags at the end of the sampling period As Businger and Oncley pointed out, the measurement

is known equally as conditional sampling or relaxed eddy accumulation (CS/REA) Figure 8 shows a schematic of a typical conditional sampling system in which fluxes of CH 4 were routinely monitored on-line In this representation, the system is set up to measure simultaneous fluxes of methane and nitrous oxide The vertical wind speed is measured by a sonic anemometer (center left of the diagram) and, dependent on whether an updraught or a downdraught has been measured, the appropriate sample valve is opened and air is sucked into a reservoir or passed into a gas analyzer Further details of the systems shown here can be found in Beverland et al., 1996 (The tunable diode laser in Fig 8 is operating as part

of a ducted eddy covariance system.) In general, the CS/REA method has been well validated over the past few years, although for fluxes of CO 2 and water vapor, eddy covariance remains the more appropriate choice (Pattey et al., 1992; Oncley et al., 1993) Conditional sampling has a role

in measuring fluxes of biogenic compounds and other trace gases for which suitable eddy covariance sensors either do not exist, e.g., nonmethane hydrocarbons (Moncrieff et al., 1998; Christensen, 2000) and herbicides (Pattey et al., 1995) or are very expensive and complex, e.g., for N 2 O and

CH 4 by tunable diode lasers (see Sec IV.D) The use of the semiempirical

beta factor worried some researchers, but it has since been put on a much sounder footing by its interpretation with the statistics of turbulence (Milne et al., 1999; Baker, 2000).

decided upon before sampling begins, and the sampling valves are only energized when

to obtain information on the scalar independently of the REA method, e.g., c0 (or a similar scalar with similar probability distribution) could be obtained by simultaneous measurements by eddy covariance Wichura et al (2000) have shown how to use hyperbolic REA to measure fluxes of the stable isotope 13 C over a spruce forest.

F Mass Balance Method

The mass balance method is well named, since it involves accounting for the mass of material leaving a well-defined volume through any of its sides Material emitted at the surface, for example CH 4 by ruminant animals within the volume, will be carried by the turbulent wind outwith the volume;

in the mass balance method, the wind speed is measured at the same heights

as the concentration of the gas species of interest, since the net horizontal flux density of a gas, q, is the time mean of the product of the instantaneous wind speed, u, and gas density  g (Denmead et al., 1998), i.e., q ¼ u g

Figure 9 shows an open enclosure that could be used to measure the gaseous emission from sheep The ‘‘fence’’ enclosing the plot is made from sample tube; air can be drawn from any or all of the sides to gas analyzers The difference between the upwind and the downwind concentrations of the relevant trace gas will show the extent by which the trace gas concentration has been enriched by addition from the animals.

The mean emission rate (F ) of gas from the volume is:

F ¼ X

Zz

0

½Uzðhg4,zi
hg2,ziÞ þVzðhg3,zi
hg1,ziÞdz

Xis horizontal distance, z is height; the boundaries of the volume are 1–4 with boundaries 1 and 2 being upwind, 3 and 4 downwind; U is the vector wind normal to horizontal boundaries 2 and 4; V is the vector wind normal

to boundaries 1 and 3 Time averages are indicated by an overbar, and thus



g4,zis for example, the gas density at height z on boundary 4 This equation