The effect of atmospheric turbulence and chamber ... · Atmospheric turbulence has been shown to exert consider-able inﬂuence on gas transport in soils (Kimball and Lemon, 1971)

Biogeosciences 9 3305ndash3322 2012wwwbiogeosciencesnet933052012doi105194bg-9-3305-2012copy Author(s) 2012 CC Attribution 30 License

Biogeosciences

The effect of atmospheric turbulence and chamber deploymentperiod on autochamber CO2 and CH4 flux measurements in anombrotrophic peatland

D Y F Lai1234 N T Roulet12 E R Humphreys5 T R Moore12 and M Dalva1

1Department of Geography McGill University 805 Sherbrooke Street West Montreal QC H3A 2K6 Canada2Global Environmental and Climate Change Centre McGill University 805 Sherbrooke Street West MontrealQC H3A 2K6 Canada3Department of Geography and Resource Management The Chinese University of Hong Kong Shatin New TerritoriesHong Kong China4Institute of Environment Energy and Sustainability The Chinese University of Hong Kong Shatin New TerritoriesHong Kong China5Department of Geography and Environmental Studies Carleton University B349 Loeb Building 1125 Colonel By DriveOttawa ON K1S 5B6 Canada

Correspondence toN T Roulet (nigelrouletmcgillca)

Received 19 December 2011 ndash Published in Biogeosciences Discuss 1 February 2012Revised 23 July 2012 ndash Accepted 7 August 2012 ndash Published 24 August 2012

Abstract Accurate quantification of soil-atmosphere gas ex-change is essential for understanding the magnitude and con-trols of greenhouse gas emissions We used an automaticclosed dynamic chamber system to measure the fluxes ofCO2 and CH4 for several years at the ombrotrophic MerBleue peatland near Ottawa Canada and found that at-mospheric turbulence and chamber deployment period hada considerable influence on the observed flux rates Witha short deployment period of 25 min CH4 flux exhibitedstrong diel patterns and both CH4 and nighttime CO2 effluxeswere highly and negatively correlated with ambient frictionvelocity as were the CO2 concentration gradients in the top20 cm of peat This suggests winds were flushing the veryporous and relatively dry near-surface peat layers and reduc-ing the belowground gas concentration gradient which thenled to flux underestimations owing to a decrease in turbu-lence inside the headspace during chamber deployment com-pared to the ambient windy conditions We found a 9 to 57 underestimate of the net biological CH4 flux at any time ofday and a 13 to 21 underestimate of nighttime CO2 ef-fluxes in highly turbulent conditions Conversely there wasevidence of an overestimation ofsim 100 of net biologicalCH4 and nighttime CO2 fluxes in calm atmospheric condi-

tions possibly due to enhanced near-surface gas concentra-tion gradient by mixing of chamber headspace air by fansThese problems were resolved by extending the deploymentperiod to 30 min After 13 min of chamber closure the fluxrate of CH4 and nighttime CO2 became constant and werenot affected by turbulence thereafter yielding a reliable es-timate of the net biological fluxes The measurement biaseswe observed likely exist to some extent in all chamber fluxmeasurements made on porous and aerated substrate suchas peatlands organic soils in tundra and forests and snow-covered surfaces but would be difficult to detect unless highfrequency semi-continuous observations were made

1 Introduction

Northern peatlands in the boreal and subarctic regions playan important role in the global carbon cycle They store a to-tal of 270 to 547 Pg of carbon (C) as peat (Turunen et al2002 Yu et al 2010) which is equivalent to 18ndash25 of theC held globally in the top 1 m of soil (Jobbagy and Jack-son 2000) though these estimates are considered low whenthe peatlands containing permafrost and at higher latitudes

Published by Copernicus Publications on behalf of the European Geosciences Union

3306 D Y F Lai et al Atmospheric turbulence and chamber deployment period

are included (McGuire et al 2009 Tarnocai et al 2009)While northern peatlands have been net sinks of C for mil-lennia natural and anthropogenic disturbances could leadto substantial release of C to the atmosphere (Turetsky etal 2002) The land-atmosphere exchange of carbon dioxide(CO2) and methane (CH4) gases largely governs the overallpeatland carbon balance (Roulet et al 2007) as well as thenet radiative forcing impacts of peatlands over time (Frolk-ing et al 2006) Accurate measurement of land-atmospheregas fluxes is needed to quantify peatland gas exchange at dif-ferent spatial-temporal scales understand the environmentalcontrols on gas fluxes and parameterize peatland and globalcarbon models (Limpens et al 2008)

Flux measurement in peatlands is often done usingchamber enclosures andor micrometeorological methodseg eddy covariance Eddy covariance gives continuousspatially-averaged gas fluxes over an ecosystem at scales of104ndash105 m2 (Scanlon and Kiely 2003) but it does not pro-vide any mechanistic detail about the flux components at afiner spatial scale (Griffis et al 2000) In contrast chambermethods measure gas concentration changes in a headspaceresulting from small well-defined areas (001ndash10 m2) Moststudies on small-scale peatland C dynamics have employedmanual chambers (eg Carroll and Crill 1997 Moore et al2002 Pelletier et al 2007) which are relatively easy andinexpensive to use but difficult to sample more frequentlythan weekly because of labour constraints Automated cham-bers thus far have not been used extensively for flux measure-ments owing to the high cost and infrastructure requirementsfor installation and operation (Savage and Davidson 2003)yet they offer great opportunity to collect flux data with amuch higher temporal resolution than manual chambers

The use of chamber enclosures is known to create someartefacts and bias in flux measurements One artefact asso-ciated with the deployment of closed chambers is the accu-mulation of gases in the chamber headspace and the subse-quent decrease in diffusive concentration gradient and mea-sured flux over time (Davidson et al 2002) This suggeststhat chamber deployment period should be kept short to min-imize the build-up of gases inside the chamber On the otherhand a short chamber deployment period increases the un-certainty of flux calculation and the difficulty in obtaininga detectable flux (Khalil et al 1998) There are also poten-tial errors in flux measurement at the beginning of deploy-ment period since chamber placement on the soil surfacecan cause a disturbance of the atmospheric interfacial layer(Hutchinson et al 2000) The duration and timing of cham-ber deployment is therefore important to obtaining accurateflux estimates

Atmospheric turbulence has been shown to exert consider-able influence on gas transport in soils (Kimball and Lemon1971) Numerous studies have observed the effects of at-mospheric turbulence on gas dynamics in soil pore spacein grassland (Flechard et al 2007) pine forest (Maier etal 2010) and black spruce boreal forest (Hirsch et al

2004) Moreover highly turbulent atmospheric conditionswere found to enhance soil gas fluxes measured by bothopen chambers in a spruce forest (Subke et al 2003) andclosed chambers in an aggrading deciduous forest (Bain etal 2005) while Schneider et al (2009) observed an over-estimation of nighttime respiration under calm conditions ina boreal peatland by closed chambers Atmospheric turbu-lence has a more pronounced effect on gas exchange fromsoils with large pore size and low moisture content owing tosignificantly greater air permeability (Kimball and Lemon1971 Conen and Smith 1998) In light of these findingsthere is a need to better understand how headspace gas con-centration is affected by turbulence and pressure disturbancesarising from chamber deployment (Kutzbach et al 2007)Autochambers have the advantage of providing a large em-pirical data set covering a range of turbulence conditions thatoccur naturally in the field for statistically robust analysis

The exchange of CO2 and CH4 at the Mer Bleue bog apeatland with very porous and aerated peat was measuredusing a closed dynamic autochamber system with a shortdeployment period of 25 min We found strong diel varia-tions in CH4 flux with significantly larger nighttime emis-sions Diel patterns in CH4 flux are not as clearly prescribedas those of CO2 that typically show net uptake during the dayand emissions at night on vegetated terrain (eg Cai et al2010) Recent studies using the eddy covariance techniquea method which is not subject to chamber-specific artefactshave reported very different diel CH4 flux patterns in peat-lands Higher CH4 flux during the daytime was observed ina boreal treed fen (Long et al 2010) as well as a Dutch peatmeadow (Hendriks et al 2010) attributable to greater tem-perature and vapour pressure gradients that enhanced bothplant-mediated convective flow as well as transpiration lossof dissolved CH4 in soil water On the other hand Baldoc-chi et al (2012) reported higher CH4 emissions at night in apeatland pasture owing to collapse of the nocturnal bound-ary layer and elongation of flux footprint to elevated CH4source areas in calm conditions In a subarctic peatland dom-inated by tall graminoids no diel cycle in CH4 flux wasseen throughout the year (Jackowicz-Korczyski et al 2010)Some studies that used autochambers on wetlands have re-ported CH4 or total hydrocarbon fluxes for time periods ofa day or more (Mastepanov et al 2008 Backstrand et al2010) making it hard to deduce the variability over shortertime periods from their results Our central research ques-tion is whether the observed diel CH4 flux pattern at MerBleue is a result of natural processes or measurement arte-facts If it is the former then biological and biogeochemi-cal reasons for the repetitive pattern of variability need to bestudied to provide an explanation If the latter modificationsto the chamber methodology may be needed to eliminate orat least greatly reduce the measurement artefacts

We hypothesize that the observed diel variability ofCH4 flux at the Mer Bleue bog is largely governed byphysical processes associated with artificial and abrupt

Biogeosciences 9 3305ndash3322 2012 wwwbiogeosciencesnet933052012

D Y F Lai et al Atmospheric turbulence and chamber deployment period 3307

changes of atmospheric turbulence by chamber deploymentTurbulence-driven pressure fluctuations can induce massflow of gas from soil pore space to the atmosphere reducinggas concentration and storage in soils (Maier et al 2010)We expect that during daytime when turbulent conditionsare dominant the reduction in diffusive concentration gra-dient in surface peat before chamber lid closure will leadto a lower transient CH4 flux measured by autochamberswhen the deployment period is short The abrupt decreasein headspace turbulence following chamber deployment willcause an increase in gas storage in the peat pore space un-til sufficient time is given for the concentration gradient toadjust to the new transport resistances For the same reasonnighttime CO2 efflux will be lower when measured in highlyturbulent conditions In calm conditions mechanical mixingof headspace air by fans can increase the near-surface con-centration gradient and lead to an immediate enhancement ofgas fluxes following chamber deployment (Hutchinson et al2000) For example Schneider et al (2009) observed non-linear increases in headspace CO2 concentration when theclosed manual chamber was deployed during calm nights atthree microsites of a boreal peatland We hypothesize that inthe calm conditions that occur during some nights chamber-induced turbulence enhances both soil CO2 and CH4 effluxesmeasured during the initial period after autochamber closurebefore an equilibrium is reached between the net rates of gasproduction in peat and gas exchange across the peat surface

The objectives of our study are to (1) investigate the re-lationship between atmospheric turbulence and autochamberCH4 and nighttime CO2 fluxes (2) examine the influence ofatmospheric turbulence on the surface peat CO2 concentra-tion gradient and (3) determine the effects of chamber de-ployment period on gas fluxes measured by autochambers

2 Methods

21 Site description

Mer Bleue peatland is a 28 km2 ombrotrophic bog locatednear Ottawa Canada (4541 N 7552 W) This region hasa cool continental climate with a mean annual temperatureof 60plusmn 08C and mean annual precipitation of 9435 mm(1971ndash2000 climate normals) (Environment Canada 2011)Field measurements were made at a dome-shaped bog inthe northwest part of this peatland Peat depth reachesabout 5 to 6 m near the centre of the bog and is shal-lower (lt 03 m) near the beaver pond margin (Roulet etal 2007) The surface of this peatland is completely cov-ered bySphagnummoss (Sphagnum magellanicumBridSphagnum capillifolium(Ehrh) HedwSphagnum angusti-folium (CEO Jensen ex Russow) CEO JensenSphag-num fallx (Klinggr) Klinggr) and the vascular plant coveris dominated by low growing ericaceous evergreen shrubs(Chamaedaphne calyculata(L) Moench Ledum groen-

landicumOederKalmia angustifoliaL) with an occasionalmix of deciduous shrubs (Vaccinium myrtilloidesMichx)sedge (Eriophorum vaginatumL) and forb (Maianthemumtrifolium (L) Sloboda)

This bog is relatively dry with a mean growing season(May to October) water table depth of 427 cm from the topof hummock over 1998ndash2008 (Teklemariam et al 2010) Ithas a hummock-lawn-hollow microtopography with a meandifference of 17 cm in elevation between the hummock andhollow surfaces (Wilson 2012) Depending on the microto-pographic location the top 10ndash30 cm peat at Mer Bleue isfibric and has a total porosity and macroporosity of greaterthan 09 and 08 respectively (Dimitrov et al 2010) Air-filled porosity of surface peat in the unsaturated zone of thisbog is between 082 and 092 (Deppe et al 2010)

22 Autochamber system

We set up 10 autochambers at the Mer Bleue bog within a15-m radius about 50 m south of the eddy covariance towerChamber locations were chosen to represent the major plantcommunities while covering a range of water table and leafarea (Table 1) Wooden boardwalks were constructed to min-imize disturbance during access to the chambers The dy-namic closed autochamber system established at this bogwas identical to the one used in a moderately rich treed fen(Cai et al 2010) a temperate Douglas-fir forest (Drewittet al 2002) a boreal aspen forest (Griffis et al 2004) andtwo boreal black spruce forests (Gaumont-Guay et al 2008Bergeron et al 2009) The autochamber consisted of a trans-parent Plexiglasreg dome fitted to a PVC cylinder with ahinged aluminum frame The near semi-spherical dome hada height of 205 cm and the PVC cylinder had an internal di-ameter of 52 cm a thickness of 1 cm and a height of 385 cmHence each autochamber covered a surface area of 021 m2The PVC cylinders were inserted 16 to 31 cm deep into thepeat leaving about 7 to 22 cm above the peat surface butbelow the height of the top of the shrubs After insertionany gaps between the outside of the cylinder and surround-ing peat were filled with surface peat taken from elsewhereat the site A partially inflated bicycle tube sealed to the topof the cylinder and a foam gasket on the aluminum flange atthe dome base ensured a good seal when the chamber wasclosed A small brushless fan within the dome mixed the airin the chamber headspace and a coiled 50-cm-long open-ended vent tube on the dome top ensured equilibration ofpressure between the inside and outside of chamber duringflux measurements

Control of the autochamber system including chamber se-lection measurement timing and data acquisition was con-trolled by a datalogger (CR23X Campbell Scientific UTUSA) A pneumatic cylinder assembly (Model BFT-173-DB Bimba Manufacturing IL USA) connected to an oil-free air compressor (Model CPFAC2600P Porter Cable TNUSA) opened and closed the chamber dome Sampling tubes

wwwbiogeosciencesnet933052012 Biogeosciences 9 3305ndash3322 2012


Table 1Vascular plant species mean water table depth (maximum and minimum depths in parentheses) and peak vascular green area index(VGA) for the 10 autochambers

Chamber Vascular Water table depth (cm) VGA

number plant speciesa 2009b 2010c (m2 mminus2)

Sedge-dominated1 Ev Vm Lg Ka minus248 (minus363minus163) minus255 (minus489minus137) 0822 Ev Cc Lg Vm Ka minus295 (minus453minus184) minus314 (minus573minus187) 1084 Ev Lg Cc Ka minus187 (minus292minus75) minus210 (minus489minus73) 1357 Ev Cc Mt Lg minus191 (minus328minus94) minus219 (minus452minus82) 113

Shrub-dominated hollow

8 Mt Lg minus194 (minus346minus88) minus189 (minus441minus39) 20910 Mt Lg minus199 (minus336minus96) minus222 (minus459minus79) 200

Shrub-dominated lawn

3 Cc Vm Mt Lg minus292 (minus419minus190) minus301 (minus543minus145) 1529 Lg Mt Ka minus240 (minus364minus154) minus266 (minus483minus127) 047

Shrub-dominated hummock

5 Cc Lg Vm minus353 (minus471minus248) minus370 (minus631minus241) 1126 Cc Vm minus372 (minus491minus260) minus380 (minus618minus263) 156

Cc Chamaedaphne calyculata Ev Eriophorum vaginatum Ka Kalmia angustifoliaLg Ledum groenlandicum Mt Maianthemum trifolium Vm Vaccinium myrtilloidesa In descending order of peak VGA Only species with peak VGAgt 01 m2 mminus2 are includedb From DOY 142 to 341c From DOY 85 to 339

(Synflex 1300 43 mm id Saint-Gobain Performance Plas-tics NJ USA) connected to the gas inlet and outlet locatedat the top of chamber dome led to a sampling manifold con-trolled by solenoid valves (Model DDBA-1BA MAC ValvesMI USA) Concentrations of CO2 and CH4 were measuredby a closed-path infrared gas analyzer (LI-6262 LI-CORNE USA) and a fast methane analyzer based on off-axisintegrated cavity output spectroscopy (DLT-100 Los GatosResearch CA USA) respectively The LI-6262 was man-ually calibrated weekly using ultra-high purity nitrogen gascontaining no CO2 (for zero) and 356 micromol molminus1 of CO2(for span) and the fast methane analyzer was calibrated withCH4 standard gas (18 and 501 micromol molminus1) once at the be-ginning of measurement season Air was circulated to the LI-6262 and back to the chambers at a flow rate of 35 l minminus1

by an AC linear pump (Model DDL Gast ManufacturingMI USA) The internal pump on the fast methane analyzersub-sampled the air stream at approximately 05 l minminus1 Thedatalogger pump and gas analyzers were all placed insidetemperature-controlled housings

The autochamber system was in operation between 22May and 7 December in 2009 and between 26 March and5 December in 2010 Prior to 6 July 2010 the autochamberswere programmed to close sequentially to measure gas fluxesfor 25 min to provide one measured flux for each chamberevery 30 min Fluxes were continuously measured between0200 and 2400 EST daily Between midnight and 0200

the system was used to estimate the effective volume of thechamber (Drewitt et al 2002) From 6 July 2010 onwardsthe chamber deployment period was increased from 25 to15 min and one flux measurement was made for each cham-ber every three hours To investigate the effect of chamberdeployment period on measured fluxes we operated only twoautochambers using extended deployment periods of 15 minfor three days (23ndash26 June) and 30 min for four days (26ndash30 June 2010) Analog outputs from the gas analyzers weresampled at 1 Hz by the datalogger averaged every 5 s anddownloaded automatically to a PC located in a hut on siteAll high frequency data were retained for flux processing

23 CO2 concentration profile system

A profile system was set up within 25 m of the autocham-bers to monitor CO2 and water vapour concentrations inthe air at seven different levels namely 255 120 20and 10 cm above the peat surface in theChamaedaphnecanopy (5 cm above peat surface) and on theSphagnumpeat surface of both a hummock and hollow (5 and 25 cmbelow theChamaedaphnecanopy respectively) Air sam-ples were drawn through sampling tubes (Synflex 130043 mm id and Bev-A-Line IV 32 mm id) into an in-frared gas analyzer (LI-6262 LI-COR NE USA) by anAC linear pump (Model SPP-15EBS-101 Gast Manufac-turing MI USA) The seven different levels were sampled



sequentially for two minutes each using solenoid valves(Model 701N13A5P Honeywell CT USA) with air beingflushed through the system in the first minute and CO2 andwater vapour concentrations measured every two secondsand then averaged in the second minute Operation of thissystem was controlled by a datalogger (CR10X CampbellScientific UT USA)

We also measured CO2 concentrations in the peat pro-file by non-dispersive infrared CO2 sensors (Model CAR-BOCAP GMM220 Vaisala Finland) that were enclosed byan expanded polytetrafluoroethylene (PTFE) membrane per-meable to gas but not water A few meters away from theair profile system we cut out peat blocks and inserted thesensors horizontally in the pit face at 5 10 20 and 40 cmdepth at a hummock and 5 and 10 cm depth at a hollow Thesensors were installed a few years before the actual measure-ments made in 2009 to minimize any possible disturbance ef-fects Thermocouples were also installed next to the sensorsto monitor peat temperature The peat blocks were carefullyreturned to the pits Peat pore space CO2 concentration andtemperature were measured at 5-min intervals and averagedevery half hour on a datalogger (CR21X Campbell Scien-tific UT USA)

24 Ancillary field measurements

We installed thermocouples into the peat at 10 cm depth in-side each of the chamber cylinders to monitor peat tem-perature at 1 Hz for the calculation of half-hourly averagesContinuous 30-min records of water table position were ob-tained with a capacitance water level probe (Model OdysseyDataflow Systems New Zealand) placed inside a perforatedABS tube (38 cm id) in the peat besides each chamberA perforated PVC tube (125 cm id) was installed next toeach ABS tube to manually measure water table position atweekly intervals in order to calibrate the capacitance proberecords

Friction velocity (ulowast) was computed every 30 min as

ulowast =

[(uprimewprime

)2+

(vprimewprime

)2]14

(1)

from 20 Hz measurements of wind velocity measured in threedimensions (u v andw) with a sonic anemometer (ModelR3-50 Gill Instruments UK) prior to rotation to a naturalwind coordinate system (Foken 2008) Friction velocity isrelated to shear stress the rate of transfer of momentumand it varies with wind speed and stability for a given sur-face roughness (Oke 1987) As part of the eddy covariancesystem for determining peatland CO2 exchange the sonicanemometer was mounted on an instrument tower at a heightof 3 m and was also used to compute cup wind speed Abovethe canopyulowast does not change with height within the con-stant flux layer of the atmosphere (Oke 1987) Within the30 cm shrub canopy we did not directly assess wind speedbut using the CO2 concentration profile results in the text we

showed that in highly turbulent conditions gusts appearedto penetrate the canopy and into the near-surface peat Inaddition we measured air temperature (Model HMP35CFCampbell Scientific Inc UT USA) 2 m above the surfaceincoming photosynthetically active radiation (PAR ModelLI-190SA LI-COR NE USA) and barometric pressure(Model CS105 Campbell Scientific Inc UT USA) at thesite every 5 s and averaged them half-hourly (Lafleur et al2003)

25 Data analysis

Mole fractions of CO2 and CH4 measured in the au-tochamber headspace were converted to mixing ratios us-ing the concurrent LI-6262 measurement of water vapourconcentration to account for water vapour dilution effectsMethane flux (micromol CH4 mminus2 hminus1) and nighttime CO2 efflux(mmol CO2 mminus2 hminus1) were then calculated from

Fgas= PV SRT A (2)

where Fgas is gas fluxP is barometric pressure (Pa)V

is chamber geometric volume (m3) S is rate of change inheadspace gas concentration (expressed as micromol molminus1 hminus1

for CH4 and mmol molminus1 hminus1 for CO2) R is universalgas constant (83144 Pa m3 molminus1 Kminus1) T is air temperature(K) andA is surface area of autochamber (m2) S is deter-mined by linear regression of gas concentration in chamberheadspace against time over a period of 15 min after dis-carding the data in the initial 45 s following chamber closureFluxes were accepted only ifr2 gt 09 (p lt 001 N = 19)except when CH4 flux measured with a 15-min deploymentperiod wasle 39 micromol mminus2 hminus1 Then ar2 threshold of 07was used due to both lower flux magnitude and sensitivity ofanalyzer signal output The use ofr2 as a filtering criterionmight underestimate the contribution of near-zero fluxes yetthe number of near-zero fluxes (plusmn036 mmol CO2 mminus2 hminus1

and plusmn26 micromol CH4 mminus2 hminus1) removed due to lowr2 wassmall (lt 5 of the whole data set for most chambers) More-over theulowast levels at which these rejected fluxes were mea-sured had a similar distribution as for all the half-hourlyulowast measured at the Mer Bleue site over a full year indi-cating that the rejection of these small number of near-zerofluxes did not bias the analysis towards certain turbulenceconditions Poor quality flux data obtained during periods ofequipment failure (eg broken tubing) and system mainte-nance were also removed from analysis The poor qualityfluxes could not be filtered out satisfactorily using a root-mean-square error (RMSE) threshold since these fluxes weremostly collected when there were issues with leakage or bro-ken tubing and typically had a small magnitude as well asa small RMSE All these quality control procedures led tothe removal of 3 to 17 of all CO2 fluxes and 2 to 28 of all CH4 fluxes collected for a given chamber NighttimeCO2 respiration was calculated only when the half-hour PARvalue was zero



We also accounted for the water dilution effects in theair profile CO2 concentration data and expressed these asmicromol CO2 molminus1 dry air CO2 concentrations measured bysensors in the peat profile were adjusted for temperature andpressure (Vaisala 2011) Half-hour CO2 concentration gra-dient in the top 20 cm surface peat of hummock was thenestimated as the slope of the linear regression between porespace CO2 concentration (measured at 0 5 10 and 20 cmbelow peat surface) and peat depth Pearson correlation anal-yses were conducted to establish relationships between gasfluxes concentration gradient and environmental variablesAll the data filtering and statistical analyses were conductedin MATLAB R2009a (MathWorks MA USA)

3 Results

31 Atmospheric turbulence and autochamber gasfluxes

Figure 1 shows the diel variability of monthly mean CH4 fluxas measured using a 25 min deployment period in 2009 fromthree autochambers located at a sedge-dominated hollow ashrub-dominated hollow and a shrub-dominated hummockThese three chambers represented the range of the micro-topography and plant communities commonly found in theMer Bleue peatland A distinct diel CH4 flux pattern wasobserved throughout the year for all chambers with loweremissions during the day than at night The difference inminimum daytime and maximum nighttime CH4 flux (seeFig 1) was greatest in the mid to late growing season be-tween July and September For instance the difference was703 micromol mminus2 hminus1 at the sedge-dominated hollow in Augustbut only 232 micromol mminus2 hminus1 in November Among the variouscollar locations the diel variation in CH4 flux was smallest atthe shrub-dominated hollow For example mean CH4 flux inSeptember decreased from a nighttime maximum to a day-time minimum by 411 at the shrub hollow compared to553 and 746 at the sedge hollow and shrub hummockrespectively (Fig 1)

Time series plot of half-hourly CH4 flux and ulowast whichcharacterizes the strength of atmospheric turbulence showsthat the two variables were tightly and inversely related(Fig 2) CH4 fluxes were higher at night for most of thedays over this two-week period whenulowast was low Howeveron DOY 206 whenulowast increased throughout the 24-h periodwithout dropping at night CH4 flux decreased consistentlyover the same period without exhibiting the usual diel CH4flux pattern It is also interesting to note that the response ofone variable to the change in the other was very quick Forinstance on DOY 204 a sharp drop inulowast in the middle of theday was followed by an immediate increase in CH4 flux ratemeasured by the autochamber in the same half hour

We further investigated the relationship betweenulowast andautochamber CH4 flux by correlation analysis using the half-

43

Fig 1 1012

1013

1014

1015

Fig 1 Diel variability of monthly mean autochamber CH4 fluxfrom (a) sedge-dominated hollow(b) shrub-dominated hollow and(c) shrub-dominated hummock in 2009 Error bar indicatesplusmn1 stan-dard error

hourly data collected between June and September 2009The data set was first stratified by air temperature in 5Cclasses to control for the effects of temperature on fluxrate Table 2 shows that CH4 fluxes from all the autocham-bers were significantly and negatively correlated withulowast

when air temperature was between 5 and 25C whilesome of the autochambers did not have significant corre-lations for the 0ndash5C and 25ndash30C classes with smaller



Table 2 Correlation coefficients (r) between friction velocity and autochamber CH4 flux stratified by air temperature in 5C classes fromJune to September 2009a

0ndash5C 5ndash10C 10ndash15C 15ndash20C 20ndash25C 25ndash30C

Chamber r N r N r N r N r N r N

Sedge-dominated

1 minus020lowast 135 minus041lowastlowast 373 minus045lowastlowast 836 minus046lowastlowast 1326 minus048lowastlowast 1033 minus022lowastlowast 2752 NS 129 minus023lowastlowast 341 minus023lowastlowast 756 minus025lowastlowast 1112 minus037lowastlowast 878 minus018lowastlowast 2424 NS 133 minus028lowastlowast 352 minus018lowastlowast 768 minus024lowastlowast 1142 minus028lowastlowast 898 NS 2417 minus036lowastlowast 119 minus051lowastlowast 313 minus050lowastlowast 769 minus055lowastlowast 1261 minus049lowastlowast 994 minus034lowastlowast 268


8 minus032lowastlowast 137 minus052lowastlowast 381 minus028lowastlowast 837 minus029lowastlowast 1329 minus024lowastlowast 1037 NS 27610 minus034lowastlowast 139 minus047lowastlowast 379 minus038lowastlowast 833 minus030lowastlowast 1330 minus035lowastlowast 1031 minus020lowastlowast 275




5 NS 130 minus023lowastlowast 278 minus014lowastlowast 625 minus021lowastlowast 1027 minus036lowastlowast 905 NS 2606 minus022lowastlowast 137 minus030lowastlowast 376 minus021lowastlowast 828 minus025lowastlowast 1321 minus029lowastlowast 1023 minus026lowastlowast 276

a r = correlation coefficientN = sample size NS = not significantlowast Significant at the 005 levellowastlowast Significant at the 001 level

44

Fig 2 1016

1017

1018 Fig 2Time series of half-hourly friction velocity and CH4 flux be-tween DOY 200 and 214 in 2009 using an autochamber at a sedge-dominated hollow as an example Other sites showed similar pat-terns though the magnitude differed among sites

sample sizes The relationships were stronger in the sedge-dominated chambers with correlation coefficients rangingfrom minus018 tominus055 Apart from using the instantaneousCH4 fluxes we also rank ordered and binned the flux databy ulowast into 20 groups (ulowast range 0ndash072 m sminus1 N = 188ndash218 for each group) to examine the relationship betweenulowast

and CH4 flux Negative relationships between the two areclearly seen from all the three differently located chambers(Fig 3) As meanulowast increased from 001 to 047 m sminus1 meanCH4 flux decreased by 765 from 204 to 48 micromol mminus2 hminus1

from the shrub hummock chamber by 710 from 967 to280 micromol mminus2 hminus1 from the sedge hollow chamber but only

by 321 from 523 to 355 micromol mminus2 hminus1 from the shrub hol-low chamber

We also conducted correlation analysis to determinewhetherulowast had a significant relationship with autochamberCO2 efflux as was the case with CH4 flux We used night-time CO2 efflux measurements for this analysis to avoid theconfounding effects of simultaneous CO2 uptake and releaseNighttime CO2 effluxes measured in all the autochamberswere significantly and negatively correlated withulowast witheven stronger relationships than CH4 fluxes (Table 3) Corre-lation coefficients obtained from sedge-dominated chamberswere betweenminus040 andminus080 while those from shrub-dominated chambers were betweenminus029 andminus078 More-over the rank ordered and binned autochamber CO2 ef-fluxes (ulowast range 0ndash050 m sminus1 N = 82ndash107 for each of the15 groups) showed a decreasing trend at all three locationsasulowast increased (Fig 4) Similar to CH4 flux we found thatthe extent of decrease in nighttime CO2 efflux with ulowast wassmallest at the shrub hollow with only a drop of 404 from99 to 59 mmol mminus2 hminus1 compared to a reduction of 542 from 144 to 66 mmol mminus2 hminus1 and of 589 from 141 to58 mmol mminus2 hminus1 at the shrub hummock and sedge hollowrespectively asulowast rose from 001 to 031ndash033 m sminus1 Therange ofulowast was smaller for CO2 than CH4 effluxes becausehighly turbulent conditions were dominant during the day-time and largely excluded from the nighttime CO2 efflux dataset



45

Fig 3 1019

1020

1021

1022

1023 Fig 3 Relationship between half-hourly friction velocity and au-tochamber CH4 flux from (a) sedge-dominated hollow(b) shrub-dominated hollow and(c) shrub-dominated hummock betweenJune and September 2009 Fluxes were rank ordered and binnedinto 20 groups by friction velocity (N = 188ndash218 for each group)Error bar indicatesplusmn1 standard error

32 Atmospheric turbulence and surface peat CO2concentration gradient

Ideally we would have liked to analyze the influence of at-mospheric turbulence on the concentrations of CH4 in thepeat profile but it is very difficult to obtain non-destructivehigh frequency samples It is possible that a membrane probelike those of Mastepanov and Christensen (2008) would becapable of continuous monitoring of subsurface CH4 con-centrations but we are unaware of this kind of probe beingtested and subsequently used in peatlands Instead we in-vestigated the influence of turbulence on gas dynamics inpeat pore space by examining the relationship between half-hourly ulowast and pore space CO2 concentration gradient in the

46

Fig 4 1024

1025

1026

1027

1028 Fig 4 Relationship between half-hourly friction velocity and au-tochamber nighttime CO2 flux from (a) sedge-dominated hollow(b) shrub-dominated hollow and(c) shrub-dominated hummockbetween June and September 2009 Fluxes were rank ordered andbinned into 15 groups by friction velocity (N = 82ndash107 for eachgroup) Error bar indicatesplusmn1 standard error

top 20 cm of peat measured between June and September2009 Figure 5 shows a general negative relationship be-tweenulowast and the rank ordered and binned (ulowast range 0ndash072 m sminus1 N = 224ndash234 for each of the 20 groups) surfacepeat CO2 concentration gradient Asulowast increased slightlyunder calm condition from 001 to 006 m sminus1 the mean CO2concentration gradient increased correspondingly from 213to 264 micromol molminus1 cmminus1 Yet asulowast further increased from006 to 047 m sminus1 we observed a consistent and substan-tial decrease in mean CO2 concentration gradient from 264to 30 micromol molminus1 cmminus1 Moreover the half-hourly surfacepeat CO2 concentration gradient was strongly and negativelycorrelated withulowast (r = minus056p lt 001N = 4490)



47

Fig 5 1029

1030

1031 Fig 5 Relationship between half-hourly friction velocity and CO2concentration gradient in the top 20 cm peat of a hummock betweenJune and September 2009 Gradients were rank ordered and binnedinto 20 groups by friction velocity (N = 224ndash234 for each group)Error bar indicatesplusmn1 standard error

33 Effects of chamber deployment period on measuredfluxes

We operated two autochambers with an extended deploymentperiod of 30 min (ie 24 lid closureschamberday) for fourdays and investigated how the fluxes derived from the rateof change in chamber headspace gas concentrations variedover 19 successive 15-min periods Figure 6a and c showthat autochamber CH4 fluxes from the sedge- and shrub-dominated lawns became relatively constant starting fromthe 9th and 7th 15-min periods respectively regardless ofwhether the flux increased or decreased during the initial pe-riod after chamber closure Nighttime CO2 efflux rates fromboth the sedge- and shrub-dominated chambers decreasedconsistently at the beginning of the measurement and thengenerally reached a stable level at approximately the 9th 15-min period (Fig 6b and d) Relative to the difference in fluxbetween the 1st and 2nd periods the mean absolute differ-ence in CH4 flux and nighttime CO2 flux between the 8thand 9th periods was only 8ndash16 and 16ndash23 respectivelyfor a given chamber Figure 7 depicts the typical changein headspace CO2 and CH4 concentrations over time dur-ing chamber deployment in calm and highly turbulent con-ditions

Figure 8 shows the time series of friction velocity andCH4 flux from the sedge-dominated chamber calculated dur-ing various 15-min periods over the 30 min of chamberdeployment between DOY 177 and 181 (ulowast range 001ndash051 m sminus1) Fluxes determined in the first 15-min periodequivalent to those obtained with the original protocol usinga 25-min deployment period still exhibited a diel patternthat was negatively correlated withulowast However for CH4fluxes determined in the 9th 15-min period and beyond dielvariability and correlation withulowast disappeared We found

48

Fig 6 1032

1033

1034

49

1035

1036

Fig 6 CH4 flux (DOY 180) and nighttime CO2 efflux (DOY 177ndash181) in 19 successive 15-min periods over 30 min of chamber de-ployment from(a b) a sedge-dominated chamber and(c d) ashrub-dominated chamber The different lines represent the repli-cate flux measurements made during the above-mentioned period



50

Fig 7 1037

1038

Fig 7 Examples of the concentration trace in the headspace of asedge-dominated chamber over 19 successive 15-min periods incalm and highly turbulent conditions for(a) CH4 and(b) CO2

that in calm conditions fluxes determined in the first 15-min period were considerably higher than those in the laterperiods In contrast during times of strong turbulence CH4fluxes obtained in the first 15-min period were much lowerthan those obtained in the 9th 15-min period and beyond

Based on the above results we assumed that fluxes deter-mined in the 9th 15-min period were no longer influencedby the atmospheric turbulence experienced prior to cham-ber closure and hence were more realistic estimates of thebiological fluxes We then estimated the degree of over- orunderestimation of biological fluxes associated with a shortdeployment period of 25 min by calculating the differencein fluxes determined in the 1st and 9th 15-min periods whenall the chambers were operated with a 15-min deploymentperiod between 6 July and 15 September 2010 The rela-tionships betweenulowast and the overestimation of CH4 andnighttime CO2 flux at hollow sites were well described byquadratic equations withr2 values between 044 and 074(Fig 9andashd) We found on average an overestimation of bi-ological CH4 flux of about 100 by autochambers in calmconditions The degree of flux overestimation gradually de-creased towards zero asulowast increased up to 02 and 04 m sminus1

at the sedge-dominated and shrub-dominated hollows re-spectively (Figs 9a and 8c) Whenulowast increased even fur-ther we observed a relatively constant underestimation ofbiological CH4 fluxes Flux underestimation when atmo-spheric turbulence was high was more pronounced at thesedge-dominated hollow (minus57 ) than the shrub-dominatedcounterpart (minus9 ) Similarly nighttime CO2 fluxes wereoverestimated at hollows by about 100 in calm conditions(Figs 9b and 8d) Asulowast increased to more than 05 m sminus1we found a 13ndash21 underestimation of nighttime CO2 fluxbased on the limited number of measurements made in windynights Meanwhile at the shrub hummock site the overesti-mation of CH4 and nighttime CO2 fluxes varied greatly fora givenulowast such that relationships withulowast were poorly de-

51

Fig 8 1039

1040

1041 Fig 8 Time series of friction velocity and CH4 flux from a sedge-dominated chamber determined from the 1st 5th 9th 13th and 17th15-min periods over 30 min of chamber deployment Note the log-arithmic scale used on the y-axis

scribed by quadratic equations with lowr2 values of 007and 002 respectively

We calculated the average overestimation of biologicalCH4 flux for groups ofulowast with a range of 01 m sminus1 increas-ing at 005 m sminus1 intervals based on the results in Fig 9When ulowast was within a range associated with an averageflux overestimation or underestimation of less than 20 CH4 fluxes collected with a 25-min deployment period wereretained in the data set Theulowast threshold for acceptingfluxes was smaller in both magnitude and range for a sedge-dominated (015ndash03 m sminus1) than a shrub-dominated hollow(025ndash06 m sminus1) This filtering resulted in a data set of 596to 3887 CH4 fluxes per chamber through the 2009 measure-ment period after removing 513 to 925 of the flux mea-surements per chamber Figure 10 shows the diel variabilityof monthly mean CH4 flux in 2009 from autochambers atthe three sites (also illustrated in Fig 1) after filtering withulowast thresholds After filtering the monthly variability of CH4fluxes from autochambers was still clear but much of the dielflux patterns shown in Fig 1 disappeared

4 Discussion

41 The influence of atmospheric turbulence andchamber deployment period on autochambergas fluxes

Friction velocity is commonly used as a measure of atmo-spheric turbulence and momentum transfer between the at-mosphere and plant canopy (Foken 2008) While we foundsignificant negative correlations betweenulowast and autocham-ber CH4 and nighttime CO2 fluxes (Tables 2 and 3) previ-ous studies have reported a positive effect of turbulence onecosystem gas fluxes Using the eddy covariance techniquenear-surface turbulence was shown to enhance ecosystem-level CH4 fluxes from a Siberian polygonal tundra with ahigh surface coverage of water bodies (Wille et al 2008) and



52

Fig 9 1042

1043

1044

1045

1046 Fig 9 Percent overestimation of autochamber CH4 and nighttime CO2 flux associated with the use of a short deployment period of 25 minas a function of friction velocity from(a b) sedge-dominated hollow(c d) shrub-dominated hollow and(e f) shrub-dominated hummockassuming that fluxes calculated based on headspace concentration change in the 9th 15-min period were more realistic

a Russian boreal peatland with a water table above the peatsurface during the snowmelt period (Gazovic et al 2010)due to turbulence-induced gas transfer across the air-waterinterface and release of gas bubbles that were adhered to thewater surface This mechanism does not occur at Mer Bleuebecause of the much lower water table Measurements withclosed dynamic chambers equipped with vents also demon-strated a higher CO2 efflux from soils of a deciduous forestand a temperate cropland as wind speed increased (Bain etal 2005 Suleau et al 2009) As strong winds blow acrossthe chamber vent tube a pressure deficit develops inside thechamber (a phenomenon known as the Venturi effect) whichis compensated for by mass flow of CO2-enriched air fromsoils into the chamber headspace (Conen and Smith 1998)Although we had no available data on chamber pressure toexclude the possibility of a Venturi effect we did not ob-

serve enhanced gas fluxes under windy conditions at the MerBleue bog

Using an open dynamic chamber that enabled changes inatmospheric pressure to be transmitted to the soil surfaceRayment and Jarvis (2000) found that atmospheric turbu-lence enhanced soil CO2 efflux from a boreal black spruceforest with stronger effect at sites with a thicker layer ofporous peat Moreover Subke et al (2003) observed a pos-itive correlation between the standard deviation of horizon-tal wind velocity and CO2 efflux rate from a spruce forestsoil by employing the same chamber system They suggestedthat the increase in soil CO2 fluxes was a result of pres-sure pumping induced by wind actions When wind passesover an uneven surface or changes in speed and directionhigh-frequency pressure fluctuations over the soil surfaceare created The presence of static horizontal pressure dif-ference can then cause a mass flow of gas horizontally and



Table 3 Correlation coefficients (r) between friction velocity and autochamber nighttime CO2 efflux stratified by air temperature in 5Cclasses from June to September 2009a

0ndash5C 5ndash10C 10ndash15C 15ndash20C 20ndash25C

Chamber r N r N r N r N r N

Sedge-dominated

1 minus070lowastlowast 134 minus074lowastlowast 278 minus071lowastlowast 453 minus073lowastlowast 440 minus080lowastlowast 862 minus040lowastlowast 127 minus048lowastlowast 255 minus040lowastlowast 403 minus044lowastlowast 369 minus077lowastlowast 774 minus041lowastlowast 126 minus051lowastlowast 242 minus044lowastlowast 393 minus049lowastlowast 371 minus052lowastlowast 697 minus057lowastlowast 102 minus063lowastlowast 216 minus060lowastlowast 390 minus058lowastlowast 423 minus067lowastlowast 84


8 minus029lowastlowast 132 minus050lowastlowast 277 minus048lowastlowast 453 minus041lowastlowast 440 minus054lowastlowast 8610 minus040lowastlowast 127 minus049lowastlowast 263 minus034lowastlowast 443 minus049lowastlowast 439 minus066lowastlowast 86





a r = correlation coefficientN = sample sizelowastlowast Significant at the 001 level

vertically in the soil and eventually out to the atmospherealong the pressure gradient (Takle et al 2004) This ex-plains the decrease in the amount of CO2 stored in the soilsof a permanent grassland (Flechard et al 2007) and pineforest (Maier et al 2010) with increasing turbulence Sim-ilarly at Mer Bleue the CO2 concentration gradient in thetop 20 cm of peat was greatly diminished under highly turbu-lent conditions compared to calm conditions (Fig 5) Hirschet al (2004) also reported a negative relationship betweenulowast

and CO2 concentration gradient in the top 5 cm litter layer ofa boreal forest but attributed this to wind flushing or directlypenetrating into the highly porous and air-filled layer near themoss surface Since the Mer Bleue bog has a low water tableand high surface peat porosity the large volume of air-filledpeat pore space may facilitate wind flushing into the surfacepeat in addition to turbulence-induced pressure pumping re-sulting in depletion of accumulated gases and reduction ofthe belowground gas concentration gradient

Although atmospheric turbulence was not expected to al-ter the closed environment inside the chamber headspace be-tween measurements (over 90 of time) the chamber wasleft open and hence the surface peat inside the collar wassubject to pressure pumping and wind flushing effects underhighly turbulent conditions With the chamber closed the dif-fusive CH4 and nighttime CO2 effluxes were small (Figs 3and 4) owing to a reduced concentration gradient caused byturbulence before chamber deployment Given the short de-ployment period of 25 min there was insufficient time for

the diffusive concentration gradient to adjust to the new tur-bulence conditions and associated transport resistances andmuch of the gases produced were stored in peat rather thanreleased into the chamber headspace Using a longer cham-ber deployment period of 30 min we show that CH4 fluxrate obtained in turbulent conditions was initially low butincreased over time and reached a stable levelsim 13 min afterchamber closure (Figs 6 and 7) The attainment of constantflux suggested that at this time the net rate of gas production(ie production minus consumption) in the peat matched thatof diffusive flux from the peat into the chamber headspaceAssuming fluxes determined in the 9th 15-min period bestrepresented the net biological gas production rates we foundunderestimations of 13ndash21 for CO2 and 9ndash57 for CH4fluxes for a given chamber associated with measurementsmade during strong turbulence (ulowast gt05 m sminus1) with a de-ployment period of 25 min (Fig 9) Kutzbach et al (2007)observed that for 20ndash40 of their peatland flux measure-ments the curves of headspace concentration evolution overtime had a concave-down shape that could not be explainedby the exponential model developed based on biophysicaltheory Our findings suggest that their measurements show-ing an increase in CO2 efflux over time might have beenmade when atmospheric turbulence was strong such that be-lowground concentration gradients were small due to windflushing andor pressure pumping before chamber deploy-ment



53

Fig 10 1047

1048

1049

1050

Fig 10 Diel variability of monthly mean autochamber CH4 fluxfrom (a) sedge-dominated hollow(b) shrub-dominated hollow and(c) shrub-dominated hummock in 2009 after filtering with thresh-olds in friction velocity Error bar indicatesplusmn1 standard error

On the other hand we observed substantial overestimationof both biological CH4 and nighttime CO2 effluxes of near100 in calm conditions whenulowast was low (Fig 9) Resultsusing an extended deployment period show that measuredfluxes were very high immediately after chamber closure and

then decreased rapidly over the first 13 min before stabilizing(Fig 6) It has been suggested that fluxes measured by closedchambers tend to decrease continuously with time as build-up of gases in the chamber headspace reduces the diffusiveconcentration gradient which in turns lowers the diffusiveflux as a feedback from the chamber to soil (Livingston etal 2005) Yet such chamber feedback was unlikely to bethe major cause of flux decrease seen at the beginning of ourmeasurements since fluxes were maintained at a relativelyconstant level after the initial period rather than exhibiting afurther decreasing trend We suggest that the large gas fluxinitially obtained was caused by chamber-induced distur-bance In calm conditions molecular diffusion dominates gastransport and a thick atmospheric interfacial layer with steepconcentration gradient develops near the peat surface As thechamber closes the fan equipped on the chamber lid and theair stream circulated through the sample tubing to the gasanalyzers facilitate mechanical air mixing in the headspaceand disrupt the interfacial layer thus instantaneously lower-ing the gas concentration just above the peat surface Thisimmediately leads to a sharp increase in gas concentrationgradient from the peat to the atmosphere and hence the in-crease in gas flux rate (Hutchinson et al 2000) Based onmanual chamber measurements in a boreal peatland Schnei-der et al (2009) found an 18ndash31 overestimation of cumu-lative seasonal nighttime ecosystem respiration if CO2 fluxesmeasured during calm conditions were included in the dataset A test of this hypothesis would be to remove the fan anddetermine fluxes over calm nights by measuring and integrat-ing the concentration profile over the height of the chamberheadspace but we have not done this experiment to date

The effects of turbulence on chamber flux measurementsvary both spatially and temporally At hollows the diel dif-ference in CH4 flux measured with a short deployment pe-riod and the underestimation of biological gas fluxes dur-ing highly turbulent conditions were both larger at sedge-dominated than shrub-dominated sitesEriophorumsedgesat the Mer Bleue bog possess aerenchymatous tissues that actas gas conduits between the soil and atmosphere (Greenup etal 2000) By generating a pressure gradient turbulence caninduce a plant-mediated convective flow of gases from deepin the root zone to the atmosphere via the aerenchyma (Arm-strong et al 1996) In addition the large pool of CH4 storedbelowground causes the sedge sites to be highly susceptibleto advective transport mechanisms (Forbrich et al 2010)As a result the pore space gas concentration gradient andtransient diffusive flux measured by autochambers in windyconditions were reduced to a much greater extent at sedge-dominated sites compared to sites dominated by shrubs lack-ing aerenchymatous tissues Atmospheric turbulence had thesmallest effect on gas fluxes measured at shrub-dominatedhollows owing to the presence of a high water table and theabsence of plant-mediated transport The smaller volume ofair-filled pore space for the full peat profile implies that ef-fective diffusivity is reduced and only a small amount of gas



enriched in CO2 and CH4 is released from soil into the at-mosphere subsequent to chamber disturbance effects in calmconditions (Schneider et al 2009) Moreover the wind flush-ingpressure pumping effect is less pronounced at hollows asthe air-filled peat layer that stores gas for potential mass flowis thinner At hummocks we found a substantial decreasein CH4 and nighttime CO2 fluxes in highly turbulent con-ditions (Figs 3 and 4) due to the presence of a thick peatlayer with high air-filled porosity However flux measure-ment bias associated with a short chamber deployment pe-riod varied considerably at hummocks even under the sameulowast (Fig 9) possibly because of the inherently high varianceof CO2 flux (Schneider et al 2009) and small magnitudeof CH4 flux Temporally the diel difference in CH4 fluxwas largest in the mid-growing season around August andSeptember because a large amount of CH4 was produced byactive methanogenesis during this period when peat temper-ature and root exudation were relatively high Consequentlythe amount of CH4 stored in peat pore space was large (Blo-dau et al 2007) thereby enhancing the effects of wind flush-ingpressure pumping and chamber-induced disturbance onflux measurement bias

42 Implications for chamber flux measurements

Closed chambers are only capable of measuring diffusive andebullition flux but not mass flow and hence cannot providea good estimate of the actual rate of peatland-atmospheregas exchange particularly in times of strong turbulence Ray-ment and Jarvis (2000) reported a 2ndash14 underestimationof soil CO2 efflux from a boreal forest when the influenceof turbulence was not taken into account in flux determina-tion Some researchers have addressed this issue by correct-ing the surface gas fluxes measured by chambers with thepore space gas storage fluxes (Hirsch et al 2004 Maier etal 2010) Yet changes in the amount of gas stored in porespace belowground depend on both present and antecedentmeteorological conditions leading to difficulty in using thestrength of atmospheric turbulence to estimate the contribu-tion of mass flow to the overall flux (Subke et al 2003) Abrief gust of wind will cause a larger change in gas storage ifit follows a calm period than if it follows a prolonged periodof strong turbulence which would have already depleted thegas stocks within the pore space Meanwhile the reductionin concentration gradient created by the gust of wind in bothcases implies that gas fluxes obtained by closed chamberswith a short deployment period will likely be low Insteadwe suggest that closed chambers can be used to estimate thenet biological gas production rate when deployed for a longenough time that the initial turbulence-related effects are nolonger observed However it is likely that the net CH4 pro-duction rate determined will be overestimated if ebullitionoccurs during a chamber measurement as gas bubbles con-taining previously generated CH4 are released into the atmo-sphere at a much greater rate than diffusion alone

Attempts have been made to compare upscaled chamberfluxes to ecosystem level fluxes measured by eddy covari-ance system in peatlands (eg Laine et al 2006 Riutta etal 2007 Forbrich et al 2011) but such comparisons shouldtake into account the potential artefacts of both chamber andeddy covariance measurements In highly turbulent condi-tions eddy covariance may measure a higher gas exchangerate than the biological flux due to enhanced gas transportfrom soil induced by pressure changes (Gu et al 2005)while closed chambers miss the mass flow component andmay measure smaller fluxes when deployed for a short pe-riod because of the transient reduction in concentration gra-dient or in some cases may measure larger fluxes due topressure pumping as winds blow over the vent tube (Conenand Smith 1998) On the other hand in calm conditionsthe eddy covariance method underestimates biological gasfluxes because of insufficient turbulent mixing presence ofdrainage flows and errors in estimating the rate of change ingas storage below the height of eddy covariance instrumen-tation (Baldocchi 2003 Aubinet 2008) while chamber de-ployment perturbs the atmospheric boundary layer and mayresult in flux overestimation Using the same autochambersystem as ours in a boreal peatland Cai et al (2010) foundhigher seasonally integrated respiration rates from chambersthan from eddy covariance which might be partly causedby chamber artefacts that could result in overestimated CO2fluxes in calm nights

Chamber deployment period is an important considera-tion when making flux measurements It is preferable to de-ploy chambers for a period as short as possible to minimizeplant stress caused by increasing temperature and build-upof moisture in the headspace over time Moreover the useof a short deployment period diminishes chamber feedbackeffects associated with reducing concentration gradients be-tween the soil and headspace (Davidson et al 2002) enablesa linear increase in headspace gas concentration for flux de-termination (Kutzbach et al 2007) and reduces the errorsof flux estimates from soils with non-uniform physical prop-erties through the profile (Venterea and Baker 2008) Fur-thermore shortening the deployment period by half in anautomated chamber system implies a doubling of the num-ber of measurements that can be made In spite of all theadvantages associated with a short deployment period weshow at the Mer Bleue bog that initial changes in headspaceconcentration after chamber closure were greatly affected byturbulence conditions prior to chamber deployment and thusshould not be used in flux calculation This is especially im-portant for the application of non-linear regression in esti-mating fluxes Although non-linear regression takes into ac-count the effects of chamber feedback it is highly sensitive tothe initial slope of headspace gas concentration versus time(Kutzbach et al 2007 Forbrich et al 2010) It should benoted that the time it takes for the measured flux to reacha stable level unaffected by pre-deployment turbulence con-ditions is probably a function of peat porosity water table



plant type and fan speed and hence should be determinedfor specific sites and chamber systems

The calculation of gas fluxes from concentration changein chamber headspace often assumes implicitly that theheadspace turbulence is constant and equal to ambient con-ditions during deployment (Kutzbach et al 2007) Yet thisassumption can hardly be met in the field as atmosphericturbulence is always changing while the turbulence inside theheadspace of a closed chamber is constant Peatland CH4 fluxis commonly quantified by deploying manual static chambersand withdrawing air samples in the headspace periodically(eg every 5 min) over the deployment period (eg 25 min)for subsequent CH4 analysis in the laboratory (eg Pelletieret al 2007 Moore et al 2011) Based on an automaticclosed dynamic chamber system measuring gas concentra-tions at 1 Hz our findings suggest that CH4 concentrationchange will be most influenced by turbulence during the ini-tial 5 min and then progressively less with time Such varia-tions in CH4 flux rates during deployment of manually sam-pled static chambers are difficult to detect because of thesmall number of data points Similarly the turbulence-relatedimpacts on CO2 efflux are hardly noticed in measurementswith manual dynamic chambers when the duration of deploy-ment is short (sim 25 min) although the infrared gas analyzertypically measures CO2 concentrations with a high tempo-ral resolution (eg Bubier et al 2003 Pelletier et al 2011)Further studies should be done to investigate the effect ofturbulence on gas concentration evolution in the headspaceof manual chambers for example by sampling air and ana-lyzing for CH4 concentration every minute for a half hourand monitoring the change in headspace CO2 concentrationin darkness over a longer deployment period This will allowus to test the assumption of a constant rate of headspace con-centration increase throughout the deployment period usedin determining CO2 and CH4 effluxes

Since fluxes are underestimated and overestimated in highand low turbulence conditions respectively closed chamberswhen deployed only for a short time are most likely mea-suring the ldquotruerdquo biological fluxes at some intermediate at-mospheric turbulence levels The friction velocity thresholdswe derived to filter the CH4 fluxes measured with a 25-min deployment period were successful in eliminating thediel pattern of CH4 flux and could be potentially used toestimate the biological CH4 flux However since the night-time CO2 fluxes were typically measured in calm condi-tions and hence overestimated a filtering approach similarto the one used for CH4 might have removed the bulk ofavailable data Instead we suggest estimating the biologicalCO2 source strength by correcting all the nighttime CO2 ef-flux data based on the empirical relationship between mea-surement bias (ie overestimation) and friction velocity(Fig 9) We do not filter or correct the daytime CO2 flux dataas carbon exchange during this period is dominated by plantuptake However as daytime periods are often fairly turbu-lent the diffusive soil respiration component of the chamber

CO2 flux is likely underestimated Based on an aggregateddata set from chamber measurements at five sites Frolkinget al (1998) reported a mid-season average ecosystem res-piration of 20 micromol mminus2 sminus1 and maximum net ecosystemCO2 exchange ofminus25 micromol mminus2 sminus1 for northern bogs Us-ing these values if chamber CO2 efflux is underestimatedby 20 during turbulent conditions then the magnitude ofmaximum daytime net ecosystem CO2 exchange measuredby closed chambers at Mer Bleue with a deployment periodof 25 min will be overestimated by 16

5 Conclusions

We demonstrated that the diel pattern of autochamber CH4flux was caused predominantly by a change in the physicalprocesses of gas transport in relation to atmospheric turbu-lence CH4 and nighttime CO2 effluxes measured with a 25-min deployment period were negatively correlated withulowastIn highly turbulent conditions underestimation of transientfluxes was caused by a combination of an artificial and sud-den decrease in headspace turbulence by chamber deploy-ment as well as a reduction of diffusive concentration gradi-ent from wind flushing andor pressure pumping as shownby our pore space CO2 concentration data in near-surfacepeat In calm conditions chamber-induced disturbance of theatmospheric interfacial layer likely steepened the gas con-centration gradient between the peat and atmosphere whichsubsequently increased the transient gas flux This measure-ment artefact associated with a difference in headspace turbu-lence before and during chamber deployment created a biasin fluxes measured in the initial 13 min after chamber closureThe constant chamber fluxes obtained after this initial periodcan be used to indicate the strength of biological source ofCO2 and CH4 as a quasi-equilibrium between the net ratesof gas production and gas exchange across the peat surfaceis likely attained Meanwhile the pore space gas dynamicsin the peat profile deserves further attention for assessing theinfluence of a large pool of gases in the deeper peat layers onthe rate of gas exchange across the peat surface

We expect turbulence-related measurement bias to bemore significant in ombrotrophic bogs than minerotrophicfens because of the presence of a lower water table andhigher air-filled peat porosity that provides an environmentmore conducive to mass flow of gases from the peat porespace A pressure pumping effect (or ventilation) has beenobserved as well in other ecosystems that have porous sub-strates for gas storage eg boreal forests with moss in theunderstory (Hirsch et al 2004) snowpack-covered mead-ows and forests (Massman and Frank 2006) and carbonateecosystems with subterranean pores and cavities (Were et al2010) Fluxes from closed chambers in settings like theseshould be interpreted with caution as they may not quan-tify the actual rate of gas exchange between soils and the at-mosphere Chamber deployment period should be carefully



selected particularly for ecosystems with porous substratessuch that headspace concentration data are collected beyondthe initial period affected by chamber artefacts The time re-quired for chamber fluxes to reach a constant level should bedetermined for each site and chamber system before actualflux measurement in the field This bias is not only limitedto closed autochambers but may be more difficult to quan-tify with a manual closed chamber system owing to the oftenpoorer temporal resolution of headspace concentration dataIf the chamber is deployed for a period too short to eliminatethis measurement artefact we suggest filtering the CH4 dataset based on friction velocity thresholds and correcting thenighttime CO2 data set based on established empirical rela-tionships between measurement bias andulowast to estimate thebiological fluxes of these gases

AcknowledgementsFunding for this research was initially pro-vided by the Fluxnet Canada Research Network supported by theNatural Sciences and Engineering Research Council of Canada(NSERC) Canadian Foundation for Climate and AtmosphericSciences (CFCAS) and BIOCAP Canada and subsequently by theCanadian Carbon Program (CFCAS) This work was also partiallyfunded by NSERC Discovery Grants to NTR and TRM TheDepartment of Geography the Global Environmental and ClimateChange Centre (GEC3) and the Graduate and Postdoctoral Studiesof McGill University are acknowledged for their generous supportto DYFL GEC3 supported by Le Fonds de recherche du Quebecndash Nature et technologies also provided support for technicalassistance by MD

Edited by U Seibt

References

Armstrong J Armstrong W Beckett P M Halder J E LytheS Holt R and Sinclair A Pathways of aeration and the mech-anisms and beneficial effects of humidity- and Venturi-inducedconvections inPhragmites australis(Cav) Trin ex Steud AquatBot 54 177ndash197 1996

Aubinet M Eddy covariance CO2 flux measurements in nocturalconditions An analysis of the problem Ecol Appl 18 1368ndash1378 2008

Backstrand K Crill P M Jackowicz-Korczynski M Mas-tepanov M Christensen T R and Bastviken D Annual car-bon gas budget for a subarctic peatland Northern Sweden Bio-geosciences 7 95ndash108doi105194bg-7-95-2010 2010

Bain W G Hutyra L Patterson D C Bright A V Daube BC Munger J W and Wofsy S C Wind-induced error in themeasurement of soil respiration using closed dynamic chambersAgric Forest Meteorol 131 225ndash232 2005

Baldocchi D D Assessing the eddy covariance technique forevaluating carbon dioxide exchange rates of ecosystems pastpresent and future Global Change Biol 9 479ndash492 2003

Baldocchi D Detto M Sonnentag O Verfaillie J Teh YA Silver W and Kelly N M The challenges of measur-ing methane fluxes and concentrations over a peatland pastureAgric Forest Meteorol 153 177ndash187 2012

Bergeron O Margolis H A and Coursolle C Forest floorcarbon exchange of a boreal black spruce forest in easternNorth America Biogeosciences 6 1849ndash1864doi105194bg-6-1849-2009 2009

Blodau C Roulet N T Heitmann T Stewart H Beer JLafleur P and Moore T R Belowground carbon turnover ina temperate ombrotrophic bog Global Biogeochem Cy 21GB1021doi1010292005GB002659 2007

Bubier J L Bhatia G Moore T R Roulet N T and LafleurP M Spatial and temporal variability in growing-season netecosystem carbon dioxide exchange at a large peatland in On-tario Canada Ecosystems 6 353ndash367 2003

Cai T Flanagan L B and Syed K H Warmer and drier con-ditions stimulate respiration more than photosynthesis in a bo-real peatland ecosystem Analysis of automatic chambers andeddy covariance measurements Plant Cell Environ 33 394ndash407 2010

Carroll P and Crill P Carbon balance of a temper-ate poor fen Global Biogeochem Cy 11 349ndash356doi10102997GB01365 1997

Conen F and Smith K A A re-examination of closed flux cham-ber methods for the measurement of trace gas emissions fromsoils to the atmosphere Eur J Soil Sci 49 701ndash707 1998

Davidson E A Savage K Verchot L V and Navarro R Min-imizing artifacts and biases in chamber-based measurements ofsoil respiration Agric Forest Meteorol 113 21ndash37 2002

Deppe M McKnight D M and Blodau C Effects of short-termdrying and irrigation on electron flow in mesocosms of a north-ern bog and an alpine fen Environ Sci Technol 44 80ndash86doi101021es901669z 2010

Dimitrov D D Grant R F Lafleur P M and Humphreys ER Modeling peat thermal regime of an ombrotrophic peatlandwith hummock-hollow microtopography Soil Sci Soc Am J74 1406ndash1425doi102136sssaj20090288 2010

Drewitt G B Black T A Nesic Z Humphreys E R Jork EM Swanson R Ethier G J Griffis T and Morgenstern KMeasuring forest floor CO2 fluxes in a Douglas-fir forest AgricForest Meteorol 110 299ndash317 2002

Environment Canada Canadian climate normals or averages 1971ndash2000 available athttpclimateweatherofficegccaclimatenormalsindexehtml last access December 2011

Flechard C R Neftel A Jocher M Ammann C Leifeld Jand Fuhrer J Temporal changes in soil pore space CO2 con-centration and storage under permanent grassland Agric ForestMeteorol 142 66ndash84 2007

Foken T Micrometeorology Springer-Verlag Berlin Heidelberg308 pp 2008

Forbrich I Kutzbach L Hormann A and Wilmking M A com-parison of linear and exponential regression for estimating dif-fusive CH4 fluxes by closed-chambers in peatlands Soil BiolBiochem 42 507ndash515 2010

Forbrich I Kutzbach L Wille C Becker T Wu J andWilmking M Cross-evaluation of measurements of peatlandmethane emissions on microform and ecosystem scales usinghigh-resolution landcover classification and source weight mod-elling Agric Forest Meteorol 151 864ndash874 2011

Frolking S E Bubier J L Moore T R Ball T BellisarioL M Bhardwaj A Carroll P Crill P M Lafleur P MMcCaughey J H Roulet N T Suyker A E Verma S B



Waddington J M and Whiting G J Relationship betweenecosystem productivity and photosynthetically active radiationfor northern peatlands Global Biogeochem Cy 12 115ndash1261998

Frolking S E Roulet N T and Fuglestvedt J How north-ern peatlands influence the Earthrsquos radiative budget Sustainedmethane emission versus sustained carbon sequestration J Geo-phys Res 111 G01008doi1010292005JG000091 2006

Gaumont-Guay D Black T A Barr A G Jassal R S andNesic Z Biophysical controls on rhizospheric and heterotrophiccomponents of soil respiration in a boreal black spruce standTree Physiol 28 161ndash171doi101093treephys2821612008

Gazovic M Kutzbach L Schreiber P Wille C and Wilmk-ing M Diurnal dynamics of CH4 from a boreal peatland duringsnowmelt Tellus B 62 133ndash139 2010

Greenup A Bradford M McNamara N Ineson P and Lee JThe role ofEriophorum vaginatumin CH4 flux from an om-brotrophic peatland Plant Soil 227 265ndash272 2000

Griffis T J Rouse W R and Waddington J M Scaling netecosystem CO2 exchange from the community to landscape-level at a subarctic fen Global Change Biol 6 459ndash473 2000

Griffis T J Black T A Gaumont-Guay D Drewitt G B NesicZ Barr A G Morgenstern K and Kljun N Seasonal varia-tion and partitioning of ecosystem respiration in a southern bo-real aspen forest Agric Forest Meteorol 125 207ndash223 2004

Gu L Falge E M Boden T Baldocchi D D Black T ASaleska S R Suni T Verma S B Vesala T Wofsy S Cand Xu L Objective threshold determination for nighttime eddyflux filtering Agric Forest Meteorol 128 179ndash197 2005

Hendriks D M D van Huissteden J and Dolman A J Multi-technique assessment of spatial and temporal variability ofmethane fluxes in a peat meadow Agric Forest Meteorol 150757ndash774 2010

Hirsch A I Trumbore S E and Goulden M L The surfaceCO2 gradient and pore-space storage flux in a high-porosity litterlayer Tellus B 56 312ndash321 2004

Hutchinson G L Livingston G P Healy R W and Striegl RG Chamber measurement of surface-atmosphere trace gas ex-change Numerical evaluation of dependence on soil interfaciallayer and sourcesink properties J Geophys Res 105 8865ndash8875doi1010291999jd901204 2000

Jackowicz-Korczyski M Christensen T R Backstrand K CrillP Friborg T Mastepanov M and Strom L Annual cycle ofmethane emission from a subarctic peatland J Geophys Res115 G02009doi1010292008JG000913 2010

Jobbagy E G and Jackson R B The vertical distribu-tion of soil organic carbon and its relation to climateand vegetation Ecol Appl 10 423ndash436doi1018901051-0761(2000)010[0423TVDOSO]20CO2 2000

Khalil M A K Rasmussen R A and Shearer M J Fluxmeasurements and sampling strategies Applications to methaneemissions from rice fields J Geophys Res 103 25211ndash252181998

Kimball B A and Lemon E R Air turbulence effects upon soilgas exchange Soil Sci Soc Am J 35 16ndash21 1971

Kutzbach L Schneider J Sachs T Giebels M NykanenH Shurpali N J Martikainen P J Alm J and Wilmk-ing M CO2 flux determination by closed-chamber methods

can be seriously biased by inappropriate application of lin-ear regression Biogeosciences 4 1005ndash1025doi105194bg-4-1005-2007 2007

Lafleur P M Roulet N T Bubier J L Frolking S and MooreT R Interannual variability in the peatland-atmosphere carbondioxide exchange at an ombrotrophic bog Global BiogeochemCy 17 1036doi1010292002GB001983 2003

Laine A Sottocornola M Kiely G Byrne K A Wilson D andTuittila E-S Estimating net ecosystem exchange in a patternedecosystem Example from blanket bog Agric Forest Meteorol138 231ndash243 2006

Limpens J Berendse F Blodau C Canadell J G FreemanC Holden J Roulet N Rydin H and Schaepman-StrubG Peatlands and the carbon cycle from local processes toglobal implications ndash a synthesis Biogeosciences 5 1475ndash1491doi105194bg-5-1475-2008 2008

Livingston G P Hutchinson G L and Spartalian KDiffusion theory improves chamber-based measurements oftrace gas emissions Geophys Res Lett 32 L24817doi1010292005GL024744 2005

Long K D Flanagan L B and Cai T Diurnal and seasonalvariation in methane emissions in a northern Canadian peatlandmeasured by eddy covariance Global Change Biol 16 2420ndash2435 2010

Maier M Schack-Kirchner H Hildebrand E E and Holst JPore-space CO2 dynamics in a deep well-aerated soil Eur JSoil Sci 61 877ndash887doi101111j1365-2389201001287x2010

Massman W J and Frank J M Advective transport of CO2 inpermeable media induced by atmospheric pressure fluctuations2 Observational evidence under snowpacks J Geophys Res111 G03005doi1010292006JG000164 2006

Mastepanov M and Christensen T R Bimembrane diffusionprobe for continuous recording of dissolved and entrapped bub-ble gas concentrations in peat Soil Biol Biochem 40 2992ndash3003 2008

Mastepanov M Sigsgaard C Dlugokencky E J Houweling SStrom L Tamstorf M P and Christensen T R Large tundramethane burst during onset of freezing Nature 456 628ndash6302008

McGuire A D Anderson L G Christensen T R Dallimore SGuo L Hayes D J Heimann M Lorenson T D Macdon-ald R W and Roulet N Sensitivity of the carbon cycle in theArctic to climate change Ecological Monographs 79 523ndash555doi10189008-20251 2009

Moore T R Bubier J L Frolking S E Lafleur P M andRoulet N T Plant biomass and production and CO2 exchangein an ombrotrophic bog J Ecol 90 25ndash36doi101046j0022-0477200100633x 2002

Moore T De Young A Bubier J Humphreys E Lafleur Pand Roulet N A multi-year record of methane flux at theMer Bleue bog southern Canada Ecosystems 14 646ndash657doi101007s10021-011-9435-9 2011

Oke T R Boundary Layer Climates 2nd Edition Routledge Lon-don 435 pp 1987

Pelletier L Moore T R Roulet N T Garneau M andBeaulieu-Audy V Methane fluxes from three peatlands in theLa Grande Riviere watershed James Bay lowland Canada JGeophys Res 112 G01018doi1010292006JG000216 2007



Pelletier L Garneau M and Moore T R Variation in CO2exchange over three summers at microform scale in a borealbog Eastmain region Quebec Canada J Geophys Res 116G03019doi1010292011JG001657 2011

Rayment M B and Jarvis P G Temporal and spatial variation ofsoil CO2 efflux in a Canadian boreal forest Soil Biol Biochem32 35ndash45 2000

Riutta T Laine J Aurela M Rinne J Vesala T Laurila THaapanala S Pihlatie M and Tuittila E-S Spatial variationin plant community functions regulates carbon gas dynamics ina boreal fen ecosystem Tellus B 59 838ndash852 2007

Roulet N T Lafleur P M Richard P J H Moore T RHumphreys E R and Bubier J L Contemporary carbon bal-ance and late Holocene carbon accumulation in a northern peat-land Global Change Biol 13 397ndash411 2007

Savage K E and Davidson E A A comparison of manual andautomated systems for soil CO2 flux measurements trade-offsbetween spatial and temporal resolution J Exp Bot 54 891ndash899 2003

Scanlon T M and Kiely G Ecosystem-scale measurements of ni-trous oxide fluxes for an intensely grazed fertilized grasslandGeophys Res Lett 30 1852doi1010292003GL0174542003

Schneider J Kutzbach L Schulz S and Wilmking M Overesti-mation of CO2 respiration fluxes by the closed chamber methodin low-turbulence nighttime conditions J Geophys Res 114G03005doi1010292008JG000909 2009

Subke J-A Reichstein M and Tenhunen J D Explaining tem-poral variation in soil CO2 efflux in a mature spruce forest inSouthern Germany Soil Biol Biochem 35 1467ndash1483 2003

Suleau M Debacq A Dehaes V and Aubinet M Wind ve-locity perturbation of soil respiration measurements using closeddynamic chambers Eur J Soil Sci 60 515ndash524 2009

Takle E S Massman W J Brandle J R Schmidt R A ZhouX Litvina I V Garcia R Doyle G and Rice C W In-fluence of high-frequency ambient pressure pumping on carbondioxide efflux from soil Agric Forest Meteorol 124 193ndash2062004

Tarnocai C Canadell J G Schuur E A G Kuhry P Mazhi-tova G and Zimov S Soil organic carbon pools in the north-ern circumpolar permafrost region Global Biogeochem Cy 23GB2023doi1010292008GB003327 2009

Teklemariam T A Lafleur P M Moore T R Roulet N Tand Humphreys E R The direct and indirect effects of inter-annual meteorological variability on ecosystem carbon dioxideexchange at a temperate ombrotrophic bog Agric Forest Mete-orol 150 1402ndash1411 2010

Turetsky M Wieder K Halsey L and Vitt D Current distur-bance and the diminishing peatland carbon sink Geophys ResLett 29 1526doi1010292001GL014000 2002

Turunen J Tomppo E Tolonen K and Reinikainen A Esti-mating carbon accumulation rates of undrained mires in Finland-application to boreal and subarctic regions The Holocene 1269ndash80doi1011910959683602hl522rp 2002

Vaisala Inc GMM220 Series CO2 Transmitter Modules UserGuide Vaisala Inc Helsinki Finland 2011

Venterea R T and Baker J M Effects of soil physical nonunifor-mity on chamber-based gas flux estimates Soil Sci Soc Am J72 1410ndash1417doi102136sssaj20080019 2008

Were A Serrano-Ortiz P Moreno de Jong C Villagarcıa LDomingo F and Kowalski A S Ventilation of subterraneanCO2 and Eddy covariance incongruities over carbonate ecosys-tems Biogeosciences 7 859ndash867doi105194bg-7-859-20102010

Wille C Kutzbach L Sachs T Wagner D and Pfeiffer E-MMethane emission from Siberian arctic polygonal tundra eddycovariance measurements and modeling Global Change Biol14 1395ndash1408 2008

Wilson P The role of patterning in spatial and temporal water ta-ble fluctuations at the Mer Bleue bog MSc Thesis GeographyDepartment McGill University Montreal 2012

Yu Z Loisel J Brosseau D P Beilman D W and Hunt SJ Global peatland dynamics since the Last Glacial MaximumGeophys Res Lett 37 L13402doi1010292010GL0435842010



are included (McGuire et al 2009 Tarnocai et al 2009)While northern peatlands have been net sinks of C for mil-lennia natural and anthropogenic disturbances could leadto substantial release of C to the atmosphere (Turetsky etal 2002) The land-atmosphere exchange of carbon dioxide(CO2) and methane (CH4) gases largely governs the overallpeatland carbon balance (Roulet et al 2007) as well as thenet radiative forcing impacts of peatlands over time (Frolk-ing et al 2006) Accurate measurement of land-atmospheregas fluxes is needed to quantify peatland gas exchange at dif-ferent spatial-temporal scales understand the environmentalcontrols on gas fluxes and parameterize peatland and globalcarbon models (Limpens et al 2008)

Flux measurement in peatlands is often done usingchamber enclosures andor micrometeorological methodseg eddy covariance Eddy covariance gives continuousspatially-averaged gas fluxes over an ecosystem at scales of104ndash105 m2 (Scanlon and Kiely 2003) but it does not pro-vide any mechanistic detail about the flux components at afiner spatial scale (Griffis et al 2000) In contrast chambermethods measure gas concentration changes in a headspaceresulting from small well-defined areas (001ndash10 m2) Moststudies on small-scale peatland C dynamics have employedmanual chambers (eg Carroll and Crill 1997 Moore et al2002 Pelletier et al 2007) which are relatively easy andinexpensive to use but difficult to sample more frequentlythan weekly because of labour constraints Automated cham-bers thus far have not been used extensively for flux measure-ments owing to the high cost and infrastructure requirementsfor installation and operation (Savage and Davidson 2003)yet they offer great opportunity to collect flux data with amuch higher temporal resolution than manual chambers

The use of chamber enclosures is known to create someartefacts and bias in flux measurements One artefact asso-ciated with the deployment of closed chambers is the accu-mulation of gases in the chamber headspace and the subse-quent decrease in diffusive concentration gradient and mea-sured flux over time (Davidson et al 2002) This suggeststhat chamber deployment period should be kept short to min-imize the build-up of gases inside the chamber On the otherhand a short chamber deployment period increases the un-certainty of flux calculation and the difficulty in obtaininga detectable flux (Khalil et al 1998) There are also poten-tial errors in flux measurement at the beginning of deploy-ment period since chamber placement on the soil surfacecan cause a disturbance of the atmospheric interfacial layer(Hutchinson et al 2000) The duration and timing of cham-ber deployment is therefore important to obtaining accurateflux estimates

Atmospheric turbulence has been shown to exert consider-able influence on gas transport in soils (Kimball and Lemon1971) Numerous studies have observed the effects of at-mospheric turbulence on gas dynamics in soil pore spacein grassland (Flechard et al 2007) pine forest (Maier etal 2010) and black spruce boreal forest (Hirsch et al

2004) Moreover highly turbulent atmospheric conditionswere found to enhance soil gas fluxes measured by bothopen chambers in a spruce forest (Subke et al 2003) andclosed chambers in an aggrading deciduous forest (Bain etal 2005) while Schneider et al (2009) observed an over-estimation of nighttime respiration under calm conditions ina boreal peatland by closed chambers Atmospheric turbu-lence has a more pronounced effect on gas exchange fromsoils with large pore size and low moisture content owing tosignificantly greater air permeability (Kimball and Lemon1971 Conen and Smith 1998) In light of these findingsthere is a need to better understand how headspace gas con-centration is affected by turbulence and pressure disturbancesarising from chamber deployment (Kutzbach et al 2007)Autochambers have the advantage of providing a large em-pirical data set covering a range of turbulence conditions thatoccur naturally in the field for statistically robust analysis

The exchange of CO2 and CH4 at the Mer Bleue bog apeatland with very porous and aerated peat was measuredusing a closed dynamic autochamber system with a shortdeployment period of 25 min We found strong diel varia-tions in CH4 flux with significantly larger nighttime emis-sions Diel patterns in CH4 flux are not as clearly prescribedas those of CO2 that typically show net uptake during the dayand emissions at night on vegetated terrain (eg Cai et al2010) Recent studies using the eddy covariance techniquea method which is not subject to chamber-specific artefactshave reported very different diel CH4 flux patterns in peat-lands Higher CH4 flux during the daytime was observed ina boreal treed fen (Long et al 2010) as well as a Dutch peatmeadow (Hendriks et al 2010) attributable to greater tem-perature and vapour pressure gradients that enhanced bothplant-mediated convective flow as well as transpiration lossof dissolved CH4 in soil water On the other hand Baldoc-chi et al (2012) reported higher CH4 emissions at night in apeatland pasture owing to collapse of the nocturnal bound-ary layer and elongation of flux footprint to elevated CH4source areas in calm conditions In a subarctic peatland dom-inated by tall graminoids no diel cycle in CH4 flux wasseen throughout the year (Jackowicz-Korczyski et al 2010)Some studies that used autochambers on wetlands have re-ported CH4 or total hydrocarbon fluxes for time periods ofa day or more (Mastepanov et al 2008 Backstrand et al2010) making it hard to deduce the variability over shortertime periods from their results Our central research ques-tion is whether the observed diel CH4 flux pattern at MerBleue is a result of natural processes or measurement arte-facts If it is the former then biological and biogeochemi-cal reasons for the repetitive pattern of variability need to bestudied to provide an explanation If the latter modificationsto the chamber methodology may be needed to eliminate orat least greatly reduce the measurement artefacts

We hypothesize that the observed diel variability ofCH4 flux at the Mer Bleue bog is largely governed byphysical processes associated with artificial and abrupt





2 Methods

21 Site description

































ulowast =

[(uprimewprime

)2+

(vprimewprime

)2]14

(1)



25 Data analysis


Fgas= PV SRT A (2)








3 Results





43

Fig 1 1012

1013

1014

1015









Sedge-dominated









44

Fig 2 1016

1017









45

Fig 3 1019

1020

1021

1022




46

Fig 4 1024

1025

1026

1027





47

Fig 5 1029

1030





48

Fig 6 1032

1033

1034

49

1035

1036




50

Fig 7 1037

1038





51

Fig 8 1039

1040




4 Discussion





52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References


















































































2 Methods

21 Site description

































ulowast =

[(uprimewprime

)2+

(vprimewprime

)2]14

(1)



25 Data analysis


Fgas= PV SRT A (2)








3 Results





43

Fig 1 1012

1013

1014

1015









Sedge-dominated









44

Fig 2 1016

1017









45

Fig 3 1019

1020

1021

1022




46

Fig 4 1024

1025

1026

1027





47

Fig 5 1029

1030





48

Fig 6 1032

1033

1034

49

1035

1036




50

Fig 7 1037

1038





51

Fig 8 1039

1040




4 Discussion





52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References








































































































ulowast =

[(uprimewprime

)2+

(vprimewprime

)2]14

(1)



25 Data analysis


Fgas= PV SRT A (2)








3 Results





43

Fig 1 1012

1013

1014

1015









Sedge-dominated









44

Fig 2 1016

1017









45

Fig 3 1019

1020

1021

1022




46

Fig 4 1024

1025

1026

1027





47

Fig 5 1029

1030





48

Fig 6 1032

1033

1034

49

1035

1036




50

Fig 7 1037

1038





51

Fig 8 1039

1040




4 Discussion





52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References





















































































ulowast =

[(uprimewprime

)2+

(vprimewprime

)2]14

(1)



25 Data analysis


Fgas= PV SRT A (2)








3 Results





43

Fig 1 1012

1013

1014

1015









Sedge-dominated









44

Fig 2 1016

1017









45

Fig 3 1019

1020

1021

1022




46

Fig 4 1024

1025

1026

1027





47

Fig 5 1029

1030





48

Fig 6 1032

1033

1034

49

1035

1036




50

Fig 7 1037

1038





51

Fig 8 1039

1040




4 Discussion





52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References

















































































3 Results





43

Fig 1 1012

1013

1014

1015









Sedge-dominated









44

Fig 2 1016

1017









45

Fig 3 1019

1020

1021

1022




46

Fig 4 1024

1025

1026

1027





47

Fig 5 1029

1030





48

Fig 6 1032

1033

1034

49

1035

1036




50

Fig 7 1037

1038





51

Fig 8 1039

1040




4 Discussion





52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References



















































































Sedge-dominated









44

Fig 2 1016

1017









45

Fig 3 1019

1020

1021

1022




46

Fig 4 1024

1025

1026

1027





47

Fig 5 1029

1030





48

Fig 6 1032

1033

1034

49

1035

1036




50

Fig 7 1037

1038





51

Fig 8 1039

1040




4 Discussion





52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References
















































































45

Fig 3 1019

1020

1021

1022




46

Fig 4 1024

1025

1026

1027





47

Fig 5 1029

1030





48

Fig 6 1032

1033

1034

49

1035

1036




50

Fig 7 1037

1038





51

Fig 8 1039

1040




4 Discussion





52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References
















































































47

Fig 5 1029

1030





48

Fig 6 1032

1033

1034

49

1035

1036




50

Fig 7 1037

1038





51

Fig 8 1039

1040




4 Discussion





52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References
















































































50

Fig 7 1037

1038





51

Fig 8 1039

1040




4 Discussion





52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References
















































































52

Fig 9 1042

1043

1044

1045










Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References



















































































Sedge-dominated















53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References
















































































53

Fig 10 1047

1048

1049

1050


















5 Conclusions







Edited by U Seibt

References



























































































5 Conclusions







Edited by U Seibt

References




















































































5 Conclusions







Edited by U Seibt

References


















































































Edited by U Seibt

References




























































































































































Documents

The effect of atmospheric turbulence and chamber ... · Atmospheric turbulence has been shown to exert consider-able inﬂuence on gas transport in soils (Kimball and Lemon, 1971)