UNIVERSITÄT BAYREUTH
Ab eilung Mik ome eo ologie
Whole-ai elaxed eddy accumula ion o he
measu emen o iso ope and ace-gas luxes
by
J. Ruppe , M. Riede e , W. A. B and and T. Foken
A bei se gebnisse
N . 51
Bay eu h, Juni 2012
2
A bei se gebnisse, Uni e si ä Bay eu h, Ab . Mik ome eo ologie, P in , ISSN 1614-8916
A bei se gebnisse, Uni e si ä Bay eu h, Ab . Mik ome eo ologie, In e ne , ISSN 1614-8924
h p://opus.ub.uni-bay eu h.de/sch i en eihen_ebene2.php?s _id=4&la=de
Eigen e lag: Uni e si ä Bay eu h, Ab . Mik ome eo ologie
Ve iel äl igung: D ucke ei de Uni e si ä Bay eu h
He ausgebe : P o . D . Thomas Foken
Monog aphien - Einzelexempla e in Baye ischen Biblio heken
Uni e si ä Bay eu h, Ab eilung Mik ome eo ologie
D-95440 Bay eu h
Die Ve an wo ung übe den Inhal lieg beim Au o .
3
Edi o ial ema ks
This epo was pa o he PhD-Thesis by J. Ruppe (2009). The pape was no accep ed because o
he a ailabili y o measu ing sys ems based on unable lase s. Ne e heless he me hod has signi ican
bene i s in ela ion o he accu acy o he measu ed luxes e en when he handling is no simple.
The e o e, he sys em was again used in he ongoing PhD-s udy by M. Riede e and upda ed in some
de ails. The ollowing epo bases on J. Ruppe ’s pape wi h some addi ional ema ks and upda es by
M. Riede e .
Th. Foken
Whole-ai elaxed eddy accumula ion o he measu emen o iso-
ope and ace-gas luxes
J. Ruppe
1,2
, M. Riede e
1
, W. A. B and
3
, T. Foken
1
1
Depa men o Mic ome eo ology, Uni e si y o Bay eu h, Ge many
2
now: Resea ch Ins i u e o he Cemen Indus y, Ve ein Deu sche Zemen we ke, Düsseldo , Ge many
3
Max Planck Ins i u e o Biogeochemis y, Jena, Ge many
Abs ac
Measu ing he iso opic composi ion o ace gas luxes can p o ide addi ional in o ma ion on ecosys em gas exchange,
when ecosys em p ocesses, like assimila ion, disc imina e agains hea ie iso opes. In he case o CO
2
exchange, di e en
mass-balances o bulk CO
2
and i s
13
CO
2
o CO
18
O iso opes can be used o sepa a e espi a ion om pho osyn he ic assimi-
la ion. Up o now, de ec o s o di ec iso ope measu emen s in he ield lack he p ecision needed o as eddy co a iance
(EC) lux measu emen s. The collec ion o upd a and downd a whole-ai samples using he elaxed eddy accumula ion
echnique (REA) allows simul aneously de e mining ace gas concen a ions and iso ope a ios by high p ecision labo a o y
analysis. A he same ime whole-ai REA elaxes se e al o he echnical p oblems ela ed o REA sampling on aps.
In es s using ai om a ank he comple e whole-ai REA sampling sys em and i s oil balloon bag ese oi s showed no
signs o con amina ion a e cleaning. The s anda d de ia ions o δ
13
C and δ
18
O iso ope a ios we e only sligh ly highe han
he p ecision speci ied o he labo a o y analysis p ocedu e. Fi s expe imen esul s showed ha iso opic di e ences (up-
d a s−downd a s) we e la ge enough o yield signal o noise a ios g ea e han i e when applying hype bolic deadbands
du ing REA sampling (HREA). The pe o mance o he ins umen and he HREA sampling me hod a e in es iga ed by
simula ion o he sampling p ocess o bulk CO
2
, which se es as p oxy scala .
Measu emen s by whole-ai HREA in combina ion wi h high p ecision iso ope analysis can quan i y he iso luxes o
13
CO
2
and CO
18
O. Fu he mo e, addi ional in o ma ion is collec ed on he scala co ela ion o bulk CO
2
and i s s able iso opes,
which ep esen s he ela i ely sho imescale o upd a s and downd a s in he u bulen exchange abo e an ecosys em.
This in o ma ion is essen ial o check he scala simila i y assump ions made in he HREA and EC/ lask me hod o he quan-
i ica ion o iso luxes.
4
1. In oduc ion
In ecen yea s, a special in e es was o quan i y he iso opic composi ion o CO
2
lux densi ies
abo e di e en ecosys ems (Bowling e al., 2003a; Ehle inge e al., 2002; Yaki and da S. L. S e n-
be g, 2000). Such measu emen s p o ide means o iden i ying he indi idual con ibu ions o sou ces
and sinks wi h di e en iso opic signa u e o he CO
2
ne ecosys em exchange (NEE) and he a e o
in e nal ecycling o CO
2
, e.g. in he canopy space o o es s (Lloyd and Fa quha , 1994; Lloyd e al.,
1996; Yaki and Wang, 1996). Two di e en mass-balances o bulk CO
2
and i s iso opes can be used
o sepa a e espi a ion om assimila ion, which disc imina es agains
13
CO
2
and CO
18
O (Bowling e
al., 2001; Wichu a e al., 2004; Wichu a, 2009).
Di e en eddy sampling me hods like elaxed eddy accumula ion (REA, Businge and Oncley,
1990) a e commonly used o measu e ace gas luxes in he bounda y laye when as high p ecision
chemical senso s a e no a ailable o eddy co a iance (EC) lux measu emen s. Eddy sampling me h-
ods a e passi e in he sense ha hey do no modi y he u bulen gas exchange o he ecosys em.
The e o e such measu emen s a e complemen a y o measu emen s wi h enclosu es e.g. on indi idual
pa s o he ecosys em and can be used o hei alida ion. Measu emen s o he u bulen exchange
abo e an ecosys em p o ide in o ma ion wi h a spa ial in eg a ion ha can close a gap o scale be-
ween iso ope s udies a lea o b anch scale and a mosphe ic iso ope s udies and la ge scale modeling
app oaches (Canadell e al., 2000; Kaplan e al., 2002; Yaki and da S. L. S e nbe g, 2000). This is
especially impo an o he in es iga ion o ca bon budge s o o es s, because in all ege a ion com-
plex gas exchange p ocesses exis .
The abili y o analyze he iso opic signa u e o he u bulen exchange is mainly limi ed by he
measu emen unce ain y ega ding he CO
2
iso ope a ios a small di e ences o bulk CO
2
mixing
a ios (Bowling e al., 1999a; Bowling e al., 1999b; Bowling e al., 2003b; Zobi z e al., 2006). Mos
s udies on iso ope lux measu emen s abo e he canopy ocus on he e alua ion o he
13
C-iso ope
signa u es. In gene al, he di e ence o iso ope signa u es in he CO
2
exchange du ing he day is ex-
pec ed o be la ge o
18
O-iso opes, because he
18
O-iso ope signa u e o CO
2
can equilib a e wi h
18
O-
deple ed soil wa e and
18
O-en iched lea wa e pools (Yaki and da S. L. S e nbe g, 2000). CO
18
O
iso ope luxes could he e o e yield mo e independen in o ma ion on assimila ion and espi a ion.
Howe e , measu emen esul s p esen ed by Bowling e al. (1999a) we e less uni o m, which migh
e lec highe empo al and spa ial a iabili y o he wa e pools.
The aim o his s udy is o p esen a me hod o he measu emen o
13
CO
2
and CO
18
O iso ope lux-
es based on he hype bolic elaxed eddy accumula ion me hod (HREA) and whole-ai sampling. The
applica ion o he hype bolic sampling c i e ia maximizes scala concen a ion di e ences (Bowling e
al., 1999b). Whole-ai sampling allows subsequen high p ecision iso ope analysis in a labo a o y di-
ec ly om he accumula ed upd a and downd a ai samples (Bowling e al., 2003a). Wi h he con-
s uc ion o a new sampling sys em we aimed a u he imp o ing he accu acy o iso ope sampling
especially o
18
O iso opes. Sample olumes we e inc eased in o de o also allow di ec and p ecise
analysis o he co esponding bulk CO
2
mixing a ios. The in eg i y o iso ope samples and sampling
accu acy was ho oughly es ed in he labo a o y and in he ield by compa ison wi h independen
measu emen s abo e a sp uce o es du ing he expe imen WALDATEM-2003 and abo e ex ensi ely
managed g assland du ing he expe imen FORKAST-TP5 in 2010.
The combined in o ma ion o CO
2
iso ope a ios and mixing a ios in upd a and downd a ai
samples is used o analyze he scala co ela ion, which is a basic assump ion in he HREA and
EC/ lask me hods (Bowling e al., 2003a). Based on he measu ed bulk CO
2
, he HREA sampling p o-
5
cedu e and lux de e mina ion me hod a e alida ed as eques ed by Bowling e al. (1999a) and K amm
e al. (1999). The me hodological pe o mance o HREA is in es iga ed by simula ion o he sampling
p ocess. Howe e , he e ec i e sampling e iciencies de e mined om measu ed bulk CO
2
da a a e
aken in o accoun o
he de e mina ion o u bulen iso luxes.
2. Theo y
In he condi ional sampling o elaxed eddy accumula ion me hod (REA) (Businge and Oncley,
1990) he u bulen lux is de e mined om he concen a ion di e ence measu ed in upd a and
downd a ai samples. This concen a ion di e ence is scaled wi h he in ensi y o u bulen e ical
mixing measu ed by he s anda d de ia ion o he e ical wind speed σ
w
based on he assump ion o
lux- a iance simila i y. Because ai sampling is no p opo ional o he e ical wind speed, i is e-
e ed o as elaxed sampling. Consequen ly, REA is an indi ec me hod o lux measu emen s. I
elies on a pa ame e iza ion in which he so called b- ac o is de e mined om a second scala quan i y
(p oxy scala ) which shows simila i y in i s a mosphe ic anspo (scala simila i y, (Ruppe e al.,
2006b; Wyngaa d and Moeng, 1992)) and o which he luc ua ions o i s concen a ion can be meas-
u ed in he ield wi h high empo al esolu ion:
(
)
c w
F b
σc c
↑ ↓
= −
. (1)
c
F
is he u bulen lux o he scala c.
c
↑
and
c
↓
a e he a e age scala concen a ions espec i ely
in upd a and downd a ai samples exp essed as d y ai mixing a ios.
Fo he p oxy scala no mally
c
F
is de e mined by eddy co a iance (EC) measu emen s
(
c
F w' c'
=
), whe e w' and c' a e he luc ua ions o he e ical wind speed w and scala concen a ion
c a ound hei a e age alues. The o e ba deno es Reynolds a e aging. The p opo ionali y ac o b
can hen be de e mined o he p oxy scala by ea anging (1) ei he based on (i) simula ion o REA
sampling on high equency scala ime se ies and i s esul ing concen a ion di e ence
c c
↑ ↓
−
o
based on (ii) measu ed concen a ion di e ence om eal REA ai sampling o he p oxy scala :
( )
w
w' c'
b
σc c
↑ ↓
=−
(2)
Many s udies demons a e he ela i e s abili y o a e age b- ac o s in uns able and mode a ely s a-
ble condi ions (Ammann and Meixne , 2002; Bake e al., 1992; Be e land e al., 1996a; Foken e al.,
1995). Ne e heless, signi ican a iabili y o b- ac o s o indi idual 30 min pe iods is obse ed, and
di e en ac o s a e discussed (Gao, 1995; Guen he e al., 1996; Oncley e al., 1993; Pa ey e al.,
1993; Ruppe e al., 2006b). Se e al s udies poin ou , ha skewness in he join equency dis ibu-
ion (JFD) o w' and c' and s uc u es in he u bulen exchange a e causing changes in b- ac o s
(Fo iadi e al., 2005; Ka ul e al., 1996; Milne e al., 2001; Ruppe e al., 2006b). The s udy by Rup-
pe e al. (2006b) a ibu es obse ed a ia ion in he scala exchange o e en s a ime scales >60 s.
This limi s he use o a unique b- ac o o all imes and asks o he de e mina ion o indi idual b-
ac o s o each sampling pe iod. Unde he assump ion o scala simila i y, he b- ac o de e mined

6
o a p oxy scala by (2) is used o de i e he u bulen lux o he scala o in e es om i s measu ed
a e age concen a ion di e ence be ween upd a and downd a REA samples
c c
↑ ↓
−
by sol ing (1).
Applica ion o a wind deadband o small e ical wind speeds, in which no samples a e aken, in-
c eases he concen a ion di e ence be ween he upd a and downd a ai accumula ion ese oi s
and he eby he ce ain y o he lux measu emen , especially i chemical senso esolu ion is a limi ing
ac o (Businge and Delany, 1990; Delany e al., 1991). A he same ime, he b- ac o dec eases wi h
he size o he deadband. The wind deadband size H
w
is no mally de ined in e e ence o he no mal-
ized e ical wind speed luc ua ions:
w
w
σ
'H
w≤. (3)
In he hype bolic elaxed eddy accumula ion me hod (HREA) he deadband ejec s no only samples
wi h small luc ua ions o he e ical wind speed w' bu also samples wi h small de ia ions om he
mean scala concen a ion, which u he inc eases he concen a ion di e ence
c c
↑ ↓
−
(Bowling e
al., 1999b; Bowling e al., 2003a):
h
cw
σσ
'' H
cw ≤. (4)
The hype bolic deadband wi h he size H
h
mus be de e mined online om a p oxy scala measu ed
wi h high empo al esolu ion, which again assumes scala simila i y. A deadband educes he e-
quency o al e swi ching du ing sampling and a he same ime, he numbe o samples used o lux
calcula ion. I also educe he sensi i i y o REA me hods o unce ain de ini ion o he mean e ical
wind speed w needed o seg ega ing samples in he up and down ese oi s (Businge and Oncley,
1990; Pa ey e al., 1993). De ails on he sampling me hod and p ocedu es used in his s udy a e de-
sc ibed in Sec ion 3. A compa ison o gene al cha ac e is ics o eddy sampling me hods like REA and
HREA and di e en sou ces o e o o lux de e mina ion a e p esen ed in a pape by Ruppe e al.
(2002).
7
3. Me hods and Ma e ial
The design o he whole-ai REA sys em goes back o he p inciples ideas o condi ional sampling
o ace gases (Businge and Oncley, 1990; Delany e al., 1991; Desja dins, 1977; Oncley e al., 1993;
Pa ey e al., 1993) and is based on a design p esen ed by Bowling e al. (2003a) in which oil balloon
bags se e as in e media e s o age o upd a and downd a ai samples a ambien p essu e.
3.1. Scala simila i y
The de e mina ion o b- ac o s in he REA me hod and he online de ini ion o a hype bolic dead-
band (i.e. in he HREA me hod) equi es he selec ion o a p oxy scala , which shows good scala
simila i y wi h he scala s o in e es . Fo his s udy he bulk CO
2
densi y signal (ρ
CO2
) o an open pa h
gas analyze was selec ed as p oxy scala o he es ima ion o he scala in ensi y o
13
C and
18
O iso-
opes
1
o CO
2
. The assump ion is, ha bulk CO
2
densi y shows su icien scala simila i y wi h he
unknown as luc ua ions o he CO
2
iso opic composi ion. A de ailed discussion o he e ec s o
scala simila i y in REA lux measu emen s is p esen ed by Ruppe e al. (2006b) and in he diploma
hesis o Hübne (2010).
We a e con iden , ha he assump ion is jus i ied a leas o he e icien selec ion o he s ong up-
and downd a s by a hype bolic deadband, as all CO
2
iso ope u bulen exchange is pa o he bulk
CO
2
u bulen exchange. Also, linea ela ionships be ween he δ
13
C iso ope a io and bulk CO
2
mixing
a io o whole-ai samples collec ed a imescales down o 500 ms and o REA samples a e epo ed by
Bowling e al. (1999a; 2001). Ne e heless, i conside ing he loca ion o sou ces and sinks in he eco-
sys em indi idually o bulk CO
2
,
13
CO
2
and CO
18
O, some di e ence in he scala exchange should be
expec ed, which migh also a ec scala simila i y. Less scala simila i y would in oduce some e o
in HREA lux esul s wi h a endency o unde es ima ing he lux (Ruppe e al., 2006b). The as-
sump ion o scala simila i y made he e is he e o e es ed by in es iga ing he δ
13
C/CO
2
and δ
18
O/CO
2
ela ions (Sec ion 4.4).
The ollowing sec ions desc ibe he implemen ed online u bulence da a analysis, he HREA sam-
pling p ocedu es, he whole-ai REA sys em design o high p ecision iso ope and ace-gas sampling
and sample analysis.
3.2. Axis o a ion and hype bolic deadband de ini ion
A h ee-dimensional plana i o a ion ma ix was de e mined based on 1 mon h o wind eloci y
da a om he sonic anemome e used o eddy co a iance and REA sampling. I indica ed good ho i-
zon al o ien a ion o he sonic anemome e and ha only mino plana - i co ec ions we e necessa y
o he e ical wind speed.
The de e mina ion o he u bulen CO
2
lux densi ies om EC measu emen s was pe o med wi h
he TK2 so wa e package (Maude and Foken, 2004) and common co ec ions and quali y con ol
measu es we e applied as ou lined by Ruppe e al. (2006a) including a WPL-co ec ion o densi y
1
13
C and
18
O iso ope a ios in his s udy e e o he iso ope composi ion o CO
2
, i.e. he a io o
13
CO
2
o CO
18
O o bulk
CO
2
. The iso ope a io is exp essed in δ-no a ion. All δ
13
C and δ
18
O alues a e epo ed ela i e o
13
C and
18
O iso opic
abundances in he in e na ional VPDB (Vienna Pee Dee Belemni e) and VPDB-CO
2
s anda ds espec i ely (CG99 scale,
see de ails in (We ne e al., 2001) and (We ne and B and, 2001)):
δ
13
C=[[(
13
C/
12
C)
sample
–(
13
C/
12
C)
VPDB
]/(
13
C/
12
C)
VPDB
]· 1000 (‰ VPDB).
δ
18
O=[[(
18
O/
16
O)
sample
–(
18
O/
16
O)
VPDB-CO2
]/(
18
O/
16
O)
VPDB-CO2
]· 1000 (‰ VPDB-CO
2
).
8
luc ua ions (Webb e al., 1980) and a plana - i o a ion (Wilczak e al., 2001) wi h a e ical wind
speed o se co ec ion o 0.032 m.
The p oblem o axis o a ion o REA was aised by Be e land e al. (1996b) and Monc ie e al.
(1998). We add essed his issue by applying he p e iously de e mined plana - i co ec ion o he
e ical wind componen online du ing HREA sampling and we e he eby able o co ec he e ical
wind speed o se . The coo dina e o a ion o he plana - i co ec ion (<3°) was applied sligh ly inco -
ec o he online da a used du ing HREA sampling. This was due o an uniden i ied azimu h o a ion
o 120° be ween he sonic anemome e aw online da a and s o ed da a. Howe e , simula ions showed,
ha his had only mino in luence on he upd a and downd a HREA sample seg ega ion unde he
condi ions o he WALDATEM-2003 expe imen . The esul ing ela i e e o o he HREA concen a-
ion di e ence due o he e oneous online plana - i o a ion was –2(±3)%. This indica es a small
unde es ima ion o he concen a ion di e ence on a e age, i.e. sligh ly educed e iciency in sampling
he maximum concen a ion di e ence. This equally applies o he p oxy scala and he scala o in e -
es . The e o e, he u bulen lux densi ies will ha dly be al e ed, i hey a e calcula ed om measu ed
e ec i e b- ac o s o he p oxy scala like in his s udy.
In gene al, he p oblem o axis o a ion and de ini ion o he e ical wind ec o o REA and
HREA can be add essed. Like desc ibed abo e he plana - i co ec ion me hod can be applied as de-
ailed axis o a ion p ocedu es o he co ec ion o he e ical wind speed w based on as online
analysis o he 3D wind da a wi h a compu e . This can be done wi hou ime lags om il e unc ions
as asked o by Monc ie e al. (1998) as long as (i) a 3D sonic anemome e is ins alled long enough
be o e REA sampling o collec a s a is ically meaning ul amoun o wind da a speci ic o he si e,
sonic anemome e and i s o ien a ion and (ii) he anemome e o ien a ion emains unchanged o REA
sampling. Bo h c i e ia can easily be me when REA sampling is pe o med a si es wi h pe manen ly
ins alled eddy co a iance measu emen sys ems, e.g. a FLUXNET si es.
Fo he online de ini ion o he hype bolic deadband du ing REA sampling acco ding o (4) he e -
ical wind speed luc ua ions w' we e de e mined om he 3D wind ec o a e applying he plana - i
co ec ion. The s anda d de ia ion o he e ical wind speed as well as he a e age and s anda d de-
ia ion o he CO
2
densi y we e con inuously ecalcula ed om he mos ecen 6 min o da a applying
a linea weigh ing unc ion by which he newes da a was a ed h ee imes mo e impo an han he
oldes da a.
3.3. Whole-ai REA sampling sys em and sampling p ocedu e
F om close o he measu emen pa h o a sonic anemome e ai is sampled h ough a 1 µm il e and
5 m o Dekabon ubing wi h polye hylene as inne wall ma e ial wi h a o al low a e o 6.6 L min
−1
,
which assu es p edominan ly u bulen low (Reynolds numbe = 2433) in he inle ube. Plumbing in
he sys em consis ed o s ainless s eel ubing and i ings. All s eel and glass ma e ial in he sys em
was ho oughly cleaned be o e assembling by h ee old insing wi h Ace one:Hexane 1:1 (nanog ade)
and subsequen hea ing. Connec ion o he REA sys em and o he glass lasks is made using quick
connec o s and ul a- o glass connec o s. Vi on® only is used as seal ma e ial also in memb ane
pumps and al es. The ai s eam is spli ed in o a bypass and a sub-sample o 3 L min
−1
(Figu e 1a).
Only he sub-sample is used o REA sampling o upd a s and downd a s. Cons an low a es
(σ ≤0.5%) wi h minimum p essu e d op as asked o by Bowling e al. (1998) and Monc ie e al.
(1998) a e achie ed by using low p essu e d op low me e s in combina ion wi h pulse-wid h pump
mo o d i e s o he adjus men o cons an pump pe o mance, ins ead o low con olle s. A
Na ion® gas-d ye is used o p e-d ying o he sample ai . Two h ee-way al es (V1, V2) di ec he
9
sample in o he en (deadband) o he bag 1 o bag 2 ese oi s acco ding o he sign o he e ical
wind speed and he size o he deadband (upd a , downd a ). Du ing ield expe imen s he de ini ion
o bag 1 and bag 2 as ese oi s o REA upd a o downd a samples was swi ched a e each sam-
pling in e al in o de o minimize any sys ema ic in luence o one sampling pa h. A hi d non-
ope a ing al e o he same kind is ins alled o assu e he same low es ic ion on all h ee low pa hs.
The ime lag (7 ms) esul ing om he sepa a ion o he sampling al es V1 and V2 and he selec ed
low a e and he al e esponse imes (10-20 ms) a e small enough in ela ion o a desi ed sampling
equency o 10 Hz (100 ms). In o de o allow o la ge sample olumes, each bag ese oi consis s
o wo 45 cm diame e Myla ® oil balloons, which a e equipped wi h s ainless s eel illing ubes,
pa ially pe o a ed, inse ed h ough he oil al e o he balloons and joined wi h a T- i ing (no
shown in Figu e 1a and 1b). An ai igh seal was achie ed by w apping s ong ubbe band a ound he
illing ube and oil al e. A e REA sampling and be o e illing in o 1 L glass lasks wi h PCTFE
s opcocks he ai om he ese oi s is u he d ied by passing h ough d ying aps illed wi h mag-
nesium pe chlo a e g anula e (Mg(ClO
4
)
2
). Backp essu e al es a he sys em ou le cons an ly main-
ain +500 hPa o e ambien p essu e in he d ying aps and glass lasks in o de o minimize po en ial
ac iona ion by adso p ion/deso p ion p ocesses a he ela i ely la ge su aces o he g anula e and
lasks.
The dead olume o he bag ese oi s, which canno be emo ed by pumping, is abou 20 mL. Ne -
e heless, all old sample in he ese oi s is emo ed e ec i ely (dilu ion >>10000:1) p io o REA
sampling by lushing he bag ese oi s wo imes wi h 10 L o d ied ai om sampling heigh h ough
he lushing uni and emp ying ia he lask- ill uni s. Du ing a hi d lushing cycle he bag ese oi s
and he wo lask- ill uni s a e hen condi ioned wi h d ied ai om sampling heigh . The bag ese -
oi s a e emp ied and condi ioning ai emains in he d ying aps and glass lasks du ing he nex 30-
40 min REA sampling p ocedu e. No mally abou 10 o 15 L o upd a and downd a ai we e col-
lec ed wi hin 30-40 min o REA sampling wi h a hype bolic deadband o H
h
=1.0. This allowed lush-
ing he 1 L glass lasks wi h 6 o 10- old olume o sample a +500 hPa o e p essu e. The illing p o-
cedu e is s opped sho ly be o e one o he bag ese oi s is emp ied by closing ac ua ed wo-way
al es on bo h sides o he glass lasks and hen manually closing he s opcocks o he lasks. Re e o
he Table 1 o de ails on indi idual sys em componen s.
The comple e sampling p ocedu e is con olled by he so wa e ‘ATEM’ (A mosphe ic Tu bulence
Exchange Measu emen s, (Ruppe , 2005)), which allows online moni o ing and au oma ed de ailed
documen a ion o each sampling p ocedu e. This so wa e also pe o ms he equi ed online analysis
o wind and scala da a du ing REA sampling o he de ini ion o he hype bolic deadband and co e-
sponding seg ega ion o upd a and downd a ai samples.
3.4. Changes in he whole-ai REA sampling sys em in 2010
In o de o a oid o e illing o he balloons la ge sample olumes we e ins alled. Two balloons
(45 cm diame e ) o each up- and downd a we e eplaced by one balloon each wi h 90 cm diame e .
The eby, he olume was enla ged by 28 L o 50 L. Also he isk o leakages in he balloons was e-
duced, because he well de elopmen can be supe ised much easie wi h ewe balloons. In addi ion
o ha , balloon holde s we e cons uc ed ha secu ed an e enly olding o he balloons du ing illing
and well de elopmen . The balloon is ixed o a massi e ci cula s eel ing - a ached o he REA hous-
ing - wi h binde clips and elas ic s aps ha i is always e enly moun ed and does no old. In ield
expe imen s he esea che mus pay a en ion o a ange he balloons wi h he wind di ec ion no o
c ea e a sail.
16
4. Resul s and Discussion
4.1. Foil balloon bag es s
The impo ance o he ma e ial selec ion and ea men o ai sample iso ope analysis is highligh ed
by es s esul s p esen ed by Schaue e al. (2003) in which hea ed s ainless s eel ubing, Vi on® seals
and polye hylene (PE) we e ound no o con amina e CO
2
iso ope samples. The iso opic in eg i y
ega ding δ
13
C o whole-ai samples du ing s o age o up o 60 min in Myla ® oil balloon bags wi h
PE as inne wall ma e ial was demons a ed by Bowling e al. (2003a). Howe e , a bias was ound o
esidence imes longe han 60 min and o δ
18
O. We he e o e pe o med simila es s wi h he same
kind o oil balloon bags. Ai sampled om one p essu ized ai ank was analyzed a e a ying esi-
dence ime in he balloon bags by high p ecision IRMS. A i s es wi h b and new balloons lushed
h ee imes wi h ai om he ank showed signi ican con amina ion o he ai samples wi h hea ie
iso opes (Figu e 2) depending on ime a e lushing. The con amina ion p esumably is he esul o he
elease o subs ances wi h ossil o igin om he balloon inne wall ma e ial PE.
Fo a second es , we ea ed he balloons by lushing hem o abou 4 days wi h ni ogen and d ied
ai and by exposing hem o in ense di ec sunligh . A e wa ds, 9 balloon bags we e i s illed and
emp ied h ee imes in he mo ning o he second es and hen illed consecu i ely wi h ai o m a ank
du ing he day, allowing di e en sample esidence imes up o 2 h be o e analysis. Only a e he
ea men , he balloon bags lacked signs o signi ican con amina ion (Figu e 3, (WALDATEM-2003
(a) and (b), FORKAST-2010 (c). The s anda d de ia ion o bo h δ
13
C and δ
18
O iso ope a ios om he
9 samples we e accep ably low and he a e age alue o he balloon bag samples compa ed well wi h
an ai sample aken di ec ly om he ank and s o ed in a glass lask. These esul s demons a e he
gene al sui abili y o oil balloon bags o bo h
13
C and
18
O iso ope ai sampling and in e media e s o -
age, which is no mally es ic ed o 30-40 min in REA.
δ13C(‰VPDB)
-9.00
-8.00
-7.00
-6.00
-5.00
-4.00
-3.00
0
2
4
6
8
δ18O(‰VPDB-CO2)
-6.50
-6.00
-5.50
-5.00
-4.50
-4.00
imea e  lushing(h)
a)
b)
Figu e 2. Foil balloon bag es o
13
C (a) and
18
O (b) iso ope sampling be o e cleaning. The symbols ep esen he measu ed
iso ope a ios in 6 indi idual balloon bags. Each bag was lushed 3 imes wi h sample ai om one ai ank on i s i s usage.
Dashed lines indica e he p og essi e con amina ion o he ai wi h hea ie iso opes a e he lushing p ocedu e.

17
ai
ank


δ13C(‰VPDB)
-8.80
-8.75
-8.70
-8.65
-8.60 all
samples
0:00
0:30
1:00
1:30
2:00
δ18O(‰VPDB-CO2)
-0.75
-0.70
-0.65
-0.60
-0.55
-0.50
-0.45
σ=
0.019
σ=
0.035
sample esidence imeinballoonbag(h)
ai
ank
all
samples
a)
b)
Figu e 3. Foil balloon bag es o
13
C (a) and
18
O (b) iso ope sampling a e bag cleaning (WALDATEM-2003 (a) and (b),
FORKAST-2010 (c). Symbols in he le mos sec ion o he igu e ep esen iso ope a ios measu ed in ai sampled om one
ank a e di e en esidence imes in 9 di e en balloon bags. E o ba s indica e he s anda d de ia ion o up o 12 epea ed
measu emen s on he same ai sample, which o m he basis o he speci ica ion o i s iso ope a io wi h high p ecision.
Dashed lines and he solid symbol in he igh sec ion o he igu e ep esen he a e age iso ope a io measu ed in he 9
balloon bags. The co esponding e o ba indica es he s anda d de ia ion o he 9 speci ied iso ope a ios om he balloon
bag samples. As e e ence, he igh mos sec ion o he igu e shows he iso opic a io o ai sampled om he ai ank in o a
glass lask wi hou esidence in a balloon bag. No e he di e ence in scales when compa ing o Figu e 2.
4.2. Whole-ai REA sys em es s
We epea ed he es s wi h he comple e whole-ai REA sys em desc ibed in Sec ion 3. Be o ehand,
small leaks in he REA sys em we e loca ed by pe o ming leak es s wi h high- acuum and emo ed.
In a hi d es o iso opic in eg i y (no igu e), samples we e di ec ed h ough di e en pa s o he
sys em (REA sampling uni , lushing uni , lask ill uni , see Figu e 1a). Small s anda d de ia ions o
δ
13
C (0.022‰) and δ
18
O (0.021‰) iso opic a ios in 15 samples assu ed, ha he su aces o o he
ma e ials in he sys em which con ac he sample (glass, s ainless s eel, aluminum, Vi on® seals,
Na ion®, Mg(ClO
4
)
2
g anula e) we e clean and no sou ce o sample con amina ion.
In a ou h es we ope a ed he comple e whole-ai REA sys em in he same way as o ield sam-
pling wi h au oma ed sampling p ocedu es a e h ee old lushing o he balloon bags, d ying aps
and glass lasks (see Sec ion 3.3) bu by d awing sample ai and ai o lushing and condi ioning om
a ank. The samples we e s o ed o abou 30 min in he balloon bags and illed in o glass lasks o
la e analysis. The s anda d de ia ion o δ
13
C (0.014‰) and δ
18
O (0.019‰) in 19 samples we e close
o he measu emen p ecision o he mass spec ome e and he a e age iso opic a ios ma ched well
wi h an ai sample om he ank s o ed in a glass lask (Figu e 4).
Be o e p oducing he sample J585, he comple e REA sys em, balloon bags and glass lasks we e
con amina ed wi h oom ai (do ed line), which p esumably had iso opic a ios deple ed by human
c)
18
b ea h compa ed o he p essu ized ai ank. The iso opic a ios o samples p oduced a e wa ds show
no sys ema ic de ia ion om he a e age, which indica es, ha he h ee old lushing p ocedu e is
e ec i e in emo ing old sample ai .
The es esul s wi h he ea ed balloon bags (9 samples) and he comple e whole-ai REA sys em
wi h he au oma ed ield sampling p ocedu e (19 samples) demons a e he iso opic in eg i y o sam-
ples aken wi h he sys em and he sys ems sui abili y o high p ecision iso ope sampling. Ne e he-
less, he sample ai esidence ime in he balloon bags should be es ic ed o he minimum needed o
REA sampling (30-40 min) and he samples should hen be ans e ed o glass lasks o he anspo
om he ield o he labo a o y.
The whole ai REA sys em es in 2010 was accomplished in a simila way. The con amina ion wi h
oom ai in he beginning o he es becomes isible in he J1 samples by sligh ly highe δ
13
C alues
(Figu e 4c). The no maliza ion o he alues J2 (and all ollowing alues) p oo s, ha he sys em is no
longe con amina ed a e only one lushing p ocess.


all
samples
δ13C(‰VPDB)
-8.80
-8.75
-8.70
-8.65
-8.60
J401
J403
J404
J405
J406
J407
J408
J409
J410
J411
J412
J413
J414
oomai
J585
J586
J587
J588
J589
J415
δ18O(‰VPDB-CO2)
-0.75
-0.70
-0.65
-0.60
-0.55
-0.50
-0.45
σ=
0.014
σ=
0.019
ai
ank
samplesinch onologicalo de
ai
ank
all
samples
a)
b)
Figu e 4. Tes o he comple e whole-ai REA sys em o
13
C (a) and
18
O (b) iso ope sampling cleaning (a) and (b) WAL-
DATEM-2003, (c) FORKAST-2010). The usage o he symbols and lines ollows he logic o Figu e 3. Ai om he ank
was used o h ee old lushing and di ec ed h ough he REA sampling sys em, s o ed in he balloon bags and sampled in o
glass lasks in he same manne as equi ed o REA sampling in he ield. Subsequen ly, he
13
C and
18
O iso ope a ios o
CO
2
sampled in o he glass lasks we e measu ed.
c)
19
4.3. HREA simula ion and b- ac o s
In o de o check he e iciency o he upd a and downd a sample seg ega ion HREA sampling
was simula ed o each sampling in e al using he ac ual al e swi ching eco d om he ield ex-
pe imen s o seg ega e and i ually accumula e upd a and downd a samples o he µ
CO2
ime se ies
de e mined by (5).
The simula ed a e age upd a and downd a mixing a ios a e compa ed o a e age upd a and
downd a CO
2
mixing a ios measu ed in he whole-ai samples in Figu e 5. The leas squa e linea
eg ession o upd a and downd a samples is well de ined (R
2
=0.93) and esul s in a slope e y
close o one. This is a p oo o co ec ins umen pe o mance ega ding he HREA sample seg ega-
ion and accumula ion p ocess. The a e age o se o he measu ed alues o +0.26 µmol mol
−1
indi-
ca es good calib a ion o ield ins umen s.
Csimula ion(µmolmol-1)
360 365 370
Cmeasu ed(µmolmol-1)
360
365
370
R2=0.93
Figu e 5. Measu ed upd a and downd a CO
2
mixing a ios in HREA whole-ai samples plo ed agains CO
2
mixing a ios
om simula ion o HREA sampling based on he EC CO
2
ime se ies (upd a solid and downd a un illed iangles) wi h
hei leas squa e linea eg ession (line).
E ec i e b- ac o s we e de e mined om he measu ed upd a and downd a CO
2
mixing a io
di e ence
C
↑
¯¯–
C
↓
¯¯ and he u bulen CO
2
lux densi y measu ed by EC acco ding o (2). The e ec i e b-
ac o s can be compa ed o b- ac o s de i ed om he simula ed mixing a io di e ences (Figu e 6).
The simula ed b- ac o s a e e y sensi i e o he applied densi y co ec ion (5), because o small mix-
ing a io di e ences ela i e o he absolu e alues o he mixing a io. Simula ions wi h simpli ied
densi y co ec ions o he CO
2
mixing a io da a esul ed in signi ican misma ches be ween he
measu ed and simula ed upd a and downd a absolu e CO
2
mixing a ios and consequen ly less co -
ela ion be ween measu ed and simula ed b- ac o s.
Bo h measu ed and simula ed alues o he b- ac o s show he a iabili y ha mus be expec ed o
HREA as well as o REA om he skewness in he JFD o he e ical wind speed and he scala and
om sampling e ec s which depend on he eddy e e sal equency (Bake e al., 1992). Like in o he
s udies, which compa e measu ed e ec i e b- ac o s o simula ed b- ac o s (Bake e al., 1992; Be e -
land e al., 1996b; McInnes e al., 1998), we ind, ha simula ed alues end o unde es ima e meas-
u ed alues especially a highe b- ac o s. High b- ac o s we e ela ed o high u bulen lux densi y
(R
2
=0.42) and educed sampling e iciency, (R
2
=0.78), i.e. sampled concen a ion di e ence pe u bu-
len lux densi y. The obse ed unde es ima ion is he esul o some ine iciency o he physical sam-
pling p ocess in sepa a ing upd a and downd a samples compa ed o he i ual ‘digi al’ sampling
in HREA simula ions (Bake e al., 1992; Be e land e al., 1996b; Lenschow and Raupach, 1991;
20
Massman, 1991; McInnes e al., 1998; Monc ie e al., 1998). Measu ed e ec i e b- ac o s esul ing
om he eal physical sampling p ocess in eg a e such de iciencies. In o de o calcula e REA u bu-
len lux densi ies om measu ed concen a ion di e ences by (1), measu ed e ec i e b- ac o s should
he e o e be p e e ed in compa ison o simula ed b- ac o s.
Vi ual sampling esul s in a sligh ly highe concen a ion di e ences and consequen ly lowe simu-
la ed b- ac o s acco ding o (2). Simula ed b- ac o s he e o e equi e alida ion when used o eplace
e ec i e b- ac o s ha could no be measu ed, and an ins umen and expe imen speci ic co ec ion
needs o be ound (Be e land e al., 1996a; McInnes e al., 1998), e.g. based on a leas squa e linea
eg ession unc ion like shown in Figu e 6. The o e es ima ion o concen a ion di e ences om i -
ual ‘digi al’ sampling o a p oxy scala in REA simula ions e alua ed o he b- ac o s wi hou co ec-
ion acco ding o (2) can cause sys ema ic unde es ima ion o lux densi ies acco ding o (1).
The size o he esiduals o he measu ed b- ac o s in Figu e 6, quan i ied wi h 0.03 by he co e-
sponding s anda d e o , ela ed o hei ange o absolu e alues o 0.15 o 0.35 p o ides an es ima e
o he a e age unce ain y o CO
2
luxes measu ed by HREA wi h a hype bolic deadband o H
h
=1.0 o
abou 10 o 20%. This quan i ies he measu emen unce ain y o he sampling sys em and me hod in
e e ence o EC lux measu emen s o a componen o which su icien analy ical p ecision is a ail-
able du ing sample analysis (signal/noise a io >10, Sec ion 4.4).
bCO2simula ion
0.1
0.2
0.3
0.4
bCO2measu ed
0.1
0.2
0.3
0.4
R2=0.74
Figu e 6. E ec i e b- ac o s de i ed om measu ed CO
2
mixing a ios in HREA whole-ai samples in ela ion o simula ed
b- ac o s (ci cles) and hei leas squa e linea eg ession (line).
Mo e in o ma ion abou HREA simula ions and b- ac o s, i.e. diu nal cycles o b, de e mined wi h
a ious p oxy scala s, is p esen ed by Hübne (2010). The e also he e ec s o po en ially w ong b-
ac o s on he REA- lux we e e alua ed.
4.4. Maximum concen a ion di e ence by HREA o iso ope analysis
The applica ion o REA wi h hype bolic deadbands, i.e. he HREA me hod (Bowling e al., 1999b),
is in ended o maximize he concen a ion di e ence o he scala o in e es , so ha i can be esol ed
wi h su icien p ecision by labo a o y analysis. Simula ions o REA (H
w
=0.6) and HREA (H
h
=1.0)
wi h he WALDATEM-2003 da a showed a concen a ion di e ence inc ease by a ac o o 1.78 (Ta-
ble 2), which is compa able o he ac o o 1.84 de i ed om esul s p esen ed by Bowling e al.
(1999b). The ac o educes o 1.73 o he WALDATEM-2003 da a o 1.65 in he wo k o Ruppe e
al. (2006b), i impe ec scala simila i y be ween he scala o in e es and he p oxy scala is consid-
e ed. Howe e , all ac o s epo ed abo e a e he esul o simula ions wi h an ideal de ini ion o he
hype bolic deadband based on scala da a om he comple e sampling in e al (lines in Figu e 7).
21
w(ms-1)
-2 -1 0 1 2
C(µmolmol-1)
355
360
365
↓
c
↑
c
Figu e 7. Plo o he e ec i e hype bolic sample selec ion o a 30 min sampling in e al (do s) ega ding e ical wind
speed w and he open pa h CO
2
mixing a io da a C om he WALDATEM-2003 expe imen , day o he yea 188, 12:00-
12:30. The sample selec ion is based on he o iginal HREA al e swi ching eco d. Bo h axes a e scaled o he 30 min a e -
age ±4 s anda d de ia ions. The solid lines indica e he ideal hype bolic deadband wi h he size H
h
=1.0 in espec o he
30 min s a is ics. The dashed lines indica e he a e age CO
2
mixing a ios o upd a (c
↑
¯¯: w>0) and downd a (c
↓
¯¯: w<0) ai
samples, esul ing om he e ec i e hype bolic sample selec ion.
Du ing he eal sampling p ocess, only his o ic scala da a is a ailable. The mean scala alue is de-
ined based on a il e unc ion and he cen e o he hype bolic deadband mo es up and down along
he scala axis. This leads o less igo ous ejec ion o samples wi h a e age scala alues (see do s
close o he cen e o Figu e 7). Consequen ly, he a e age scala concen a ion di e ence be ween
upd a s and downd a s is sligh ly dec eased, e.g. he CO
2
mixing a io di e ence
C
↑
¯¯–
C
↓
¯¯ in Figu e 7.
The applica ion o a wind-deadband (REA) ins ead o a hype bolic deadband (HREA) would inco po-
a e e en mo e ai wi h a e age CO
2
mixing a ios (e.g. a ound 360 µmol mol
−1
in Figu e 7) in upd a
and downd a ai samples and he eby u he dec ease he mixing a io di e ence. Simula ions based
on he eco ded al e swi ching om he WALDATEM-2003 expe imen showed, ha he ealis ic
concen a ion di e ence inc ease be ween REA (H
w
=0.6) and HREA (H
h
=1.0) is only 1.63 (Table 2).
The co esponding expec ed CO
2
mixing a io di e ence om HREA simula ions based on he e ec-
i e sample seg ega ion du ing he expe imen wi h a hype bolic deadband o H
h
=1.0 is
2.4(±0.5) µmol mol
−1
. The obse ed mixing a io di e ences we e only sligh ly smalle ,
(2.3(±0.6) µmol mol
−1
, Table 2, Figu e 8a), e lec ing also he physical ai sampling e ec s discussed
in Sec ion 4.4. Howe e , hese also a ec REA sampling wi h a wind-deadband and an e ec i e ela-
i e concen a ion di e ence inc ease close o 1.63 can be assumed.

22
CO2(µmolmol-1)
-4
-3
-2
-1
0
δ13C(‰VPDB)
0.00
0.05
0.10
0.15
0.20
dayo  heyea (CET)
δ18O(‰VPDB-CO2)
-0.1
0.0
0.1
0.2
0.3
188 189 190
c)
b)
a)
5
σ
10
σ
5
σ
10
σ
Figu e 8. CO
2
mixing a io di e ences (a) and
13
C (b) and
18
O (c) iso ope a io di e ences in upd a and downd a REA
ai samples (solid diamonds).
18
O iso ope a io di e ences p esumably in luenced by incomple e d ying ou o he pe iod
ma ked wi h a g ay ba a e indica ed as un illed diamonds. Do ed lines indica e he i e old and en old s anda d de ia ions
speci ied in Sec ion 4.2 and Figu e 4 in o de o assess he measu emen p ecision o he iso ope samples.
We a oided inc easing he hype bolic deadband size H
h
u he , because in con as o esul s o ide-
al simula ions, no signi ican addi ional concen a ion di e ence inc ease could be expec ed wi h la -
ge hype bolic deadbands when acknowledging he ealis ic sampling p ocess. Also, he ep esen a-
i eness o upd a and downd a samples o he JFD would be u he dec eased i la ge deadbands
would be applied.
Du ing he WALDATEM-2003 expe imen , he o e all p opo ions o e alua ed samples in each
sampling in e al we e 14.7(±3.0)% upd a s and 9.1(±2.6)% downd a s. The di e ence in he num-
be o upd a and downd a samples esul s om skewness in he JFD and was discussed by Bowling
e al. (1999b). The au ho s sugges an op imum deadband size a H
h
=1.1 and an asymme ic adjus -
men o he hype bolas. We decided no o adjus he symme ic shape o he hype bolic deadband, in
o de o p e en a i ac s ha could esul om changing shapes o he JFD, which ha e o be consid-
e ed abo e all ege a ion.
CO
2
mixing a io di e ences o whole-ai HREA upd a and downd a samples in he ange o –
1.3 o –3.9 µmol mol
−1
exceeded he en old measu emen p ecision. Two samples ma ked wi h a ‘+’
in Figu e 5a showed ela i ely small sample p opo ions. Thei ep esen a i eness in espec o he
JFD emains ques ionable. Fu he mo e, educed p ecision o he CO
2
mixing a io di e ence can
23
esul om he ans e o ela i ely small amoun s o sample ai . The e o e, only simula ed and co -
ec ed b- ac o s we e used o he es ima ion o he iso ope luxes o hese wo samples in Figu e 8b
and 8c.
The
13
C and
18
O iso opic di e ences, i.e. he di e ence o he iso ope a ios o upd a and down-
d a ai samples, obse ed by he HREA measu emen s du ing WALDATEM-2003 we e on a e age
0.11(±0.03)‰ o δ
13
C and 0.11(±0.02)‰ o δ
18
O du ing he ime wi h su icien sample d ying (Ta-
ble 2). Mos alues lie be ween he i e old and en old s anda d de ia ions ound in he whole-ai
REA sys em es s (dashed lines in Figu e 8b and 8c, compa e Figu e 4). Simila δ
13
C di e ences we e
epo ed o many samples by Bowling e al. (1999a). Ne e heless, he small di e ences o he iso-
ope a ios ask o high p ecision in he sample analysis. Based on he compa ison o he iso opic di -
e ences and he p ecision de e mined in he whole-ai REA sys em es s a measu emen unce ain y
due o he esolu ion o he iso opic di e ences o 10% o 20% can be es ima ed.
The mixing a io di e ences obse ed o CH
4
and N
2
O in day ime samples aken abo e he sp uce
o es a Walds ein/Weidenb unnen du ing he WALDATEM-2003 expe imen anged om −2.9 o
1.6 nmol mol
−1
o CH
4
and −0.37 o 0.49 nmol mol
−1
o N
2
O. Mos mixing a io di e ences we e in
he o de o measu emen p ecision (1.3 nmol mol
−1
o CH
4
and 0.13 nmol mol
−1
o N
2
O) and conse-
quen ly oo small o be esol ed by HREA sampling and whole-ai analysis wi hou p e-concen a ion
on a ap. The indica ion o nega i e concen a ion di e ences and downwa d di ec ion o CH
4
and
N
2
O luxes on a e age o he day ime samples was no signi ican (Table 2).
Table 2. Concen a ion di e ence inc ease achie ed by HREA sampling du ing he WALDATEM-2003
expe imen and compa able da a.
Scena io Concen a ion di e ence inc ease
HREA (H
h
=1.0) / REA (H
w
=0.6)
A e age scala concen a ion
di e ences HREA (H
h
=1.0)
Simula ion, ideal 1.78 his s udy,
1.84 (Bowling e al., 1999b) CO
2
: 4.9(±2.4) µmol mol
−1
(Bowl-
ing e al., 1999b) based on da a
om Eas e n USA deciduous
o es
Simula ion, impe ec
scala simila i y
1.73 his s udy,
1.65 (Ruppe e al., 2006b)
Simula ion, p oxy
scala and σ
w
s a is ics
de ined om p e ious
da a and il e unc ion
1.63 his s udy CO
2
: 2.4(±0.5) µmol mol
−1
his s udy
Measu ed, including
physical sampling
e ec s
CO
2
: 2.3(±0.6) µmol mol
−1
δ
13
C: 0.11(±0.03) ‰ VPDB
δ
18
O: 0.11(±0.02) ‰ VPDB-CO
2
CH
4
: −0.4(±1.2) nmol mol
−1
N
2
O: −0.02(±0.26) nmol mol
−1
24
5. Conclusion
Lab expe imen s wi h oil balloon bags and he comple e whole-ai REA sys em demons a e hei
sui abili y o high p ecision iso ope sampling o bo h
13
C and
18
O iso opes o CO
2
. This was indi-
ca ed by a close ma ch o iso ope a ios ound in ai samples om wo independen sampling sys ems.
We he e o e conclude ha oil balloons a e sui able lexible ai collec ion con aine s o in e media e
s o age a e cleaning as desc ibed in Sec ion 4.1. La ge whole-ai sample olumes, p ecise low and
p essu e con ol, ca e ul ma e ial selec ion and ea men and e ec i e sample d ying helped o in-
c ease he sampling accu acy o he comple e whole-ai REA sys em especially o
18
O iso opes.
δ
13
C/CO
2
and also δ
18
O/CO
2
co ela ions can he e o e eadily be in es iga ed e en a ela i ely
small anges o CO
2
mixing a ios.
HREA measu emen s p o ide addi ional in o ma ion on he scala a ia ion and on he mos signi i-
can e en s in he u bulen iso opic exchange abo e he ecosys em o he de e mina ion o iso luxes.
The compa ison o measu emen esul s and simula ions o HREA sampling o bulk CO
2
con i med
good ins umen pe o mance and indica ed 10 o 20% unce ain y o he quan i ica ion o luxes due
o he sampling me hod. The measu ed e ec i e b- ac o s should be p e e ed o lux de e mina ion.
Simula ed b- ac o s equi e alida ion and po en ially co ec ion in o de o p e en he isk o sys em-
a ic unde es ima ion o luxes. De ailed axis o a ion p ocedu es o REA and HREA sampling wi hou
a ime lag (Be e land e al., 1996b; McInnes e al., 1998; Monc ie e al., 1998) can be implemen ed
by e alua ing he 3D wind ec o and pe o ming online plana - i co ec ions (Wilczak e al., 2001).
The p ecise synch oniza ion o REA seg ega ion al e swi ching can be achie ed unde de ined sam-
ple low condi ions by di e en ial c oss-co ela ion measu emen s as ou lined in Sec ion 3.4. A con-
cen a ion di e ence inc ease o 63% was achie ed by applying he HREA sampling me hod ins ead
o classical REA. Ne e heless, ela i ely small iso opic di e ences in upd a and downd a samples
collec ed du ing he WALDATEM-2003 and FORKAST-2010 expe imen s equi ed high p ecision
iso ope analysis. The measu emen unce ain y due o he chemical esolu ion o he iso ope a io di -
e ences was es ima ed a 10 o 20%.
Whole-ai HREA in combina ion wi h high p ecision iso ope analysis can quan i y iso luxes o
13
CO
2
and CO
18
O and collec addi ional in o ma ion on he scala co ela ion o bulk CO
2
, ep esen ing
he ela i ely sho imescale o up- and downd a s in he u bulen exchange abo e an ecosys em.
6. Acknowledgmen s
The au ho s wish o hank An hony C. Delany, Da e R. Bowling, and Nina Buchmann o help ul
commen s and discussions on he ealiza ion o he iso ope REA echnique by using whole-ai balloon
bag ese oi s. We acknowledge he suppo du ing he ield expe imen s by Ch is oph Thomas, Ma -
hias Maude , Te esa Be olini, Johannes Olesch, Johannes Lüe s and he echnical suppo pe o med
by he s a o he Bay eu h Ins i u e o Te es ial Ecosys em Resea ch (BITÖK) o he Uni e si y o
Bay eu h. The high p ecision labo a o y analysis was pe o med by Michael Ro he and A min Jo dan
in he Iso ope- and Gas labo a o y o he Max-Planck Ins i u e in Jena. This s udy was suppo ed by
he Ge man Fede al Minis y o Educa ion and Resea ch (PT BEO51-0339476 D) and he Ba a ian
S a e Minis y o Sciences, Resea ch and A s wi h he join esea ch p ojec “FORKAST” (“In es i-
ga ion o ca bon u no e o g asslands in a no he n Ba a ian low moun ain ange unde ex eme
clima e condi ions”.
25
Re e ences
Ammann, C. (1999), On he applicabili y o elaxed eddy accumula ion and common me hods o measu ing ace gas luxes,
Geog aphisches Ins i u de ETH Zü ich ed., ETH Zü ich, Zü ich, 230 pp.
Ammann, C., and F. X. Meixne (2002), S abili y dependence o he elaxed eddy accumula ion coe icien o a ious scala
quan i ies, J. Geophys. Res., 107, (D8), 4071, doi:10.1029/2001JD000649.
Bake , J. M., J. M. No man, and W. L. Bland (1992), Field-scale applica ion o lux measu emen by condi ional sampling,
Ag ic. Fo . Me eo ol., 62, 31-52.
Be e land, I. J., J. B. Monc ie , D. H. O'Neill, K. J. Ha g ea es, and R. Milne (1996a), Measu emen s o me hane and ca -
bon dioxide luxes om pea land ecosys ems by he condi ional-sampling echnique, Qua . J. Roy. Me eo . Soc., 122,
819-838.
Be e land, I. J., D. H. Oneill, S. L. Sco , and J. B. Monc ie (1996b), Design, cons uc ion and ope a ion o lux measu e-
men sys ems using he condi ional sampling echnique, A mos. En i on., 30, 3209-3220.
Bowling, D. R., A. A. Tu nipseed, A. C. Delany, D. D. Baldocchi, J. P. G eenbe g, and R. K. Monson (1998), The use o
elaxed eddy accumula ion o measu e biosphe e-a mosphe e exchange o isop ene and o he biological ace gases,
Oecologia, 116, 306-315.
Bowling, D. R., D. D. Baldocchi, and R. K. Monson (1999a), Dynamics o iso opic exchange o ca bon dioxide in a Tennes-
see deciduous o es , Glob. Biogeochem. Cycles, 13, 903-922.
Bowling, D. R., A. C. Delany, A. A. Tu nipseed, D. D. Baldocchi, and R. K. Monson (1999b), Modi ica ion o he elaxed
eddy accumula ion echnique o maximize measu ed scala mixing a io di e ences in upd a s and downd a s, J. Geo-
phys. Res., 104, (D8), 9121-9133.
Bowling, D. R., P. P. Tans, and R. K. Monson (2001), Pa i ioning ne ecosys em ca bon exchange wi h iso opic luxes o
CO
2
, Global Change Biol., 7, 127-145.
Bowling, D. R., D. E. Pa aki, and J. R. Ehle inge (2003a), C i ical e alua ion o mic ome eo ological me hods o measu ing
ecosys em-a mosphe e iso opic exchange o CO
2
, Ag ic. Fo . Me eo ol., 116, 159-179.
Bowling, D. R., S. D. Sa gen , B. D. Tanne , and J. R. Ehle inge (2003b), Tunable diode lase abso p ion spec oscopy o
s able iso ope sudies o ecosys em-a mosphe e CO
2
exchange, Ag ic. Fo . Me eo ol., 118, 1-19.
B and, W. A. (2005), O
2
/N
2
S o age Aspec s and Open Spli Mass Spec ome ic De e mina ion, in P oceedings o he 12
h
WMO/IAEA Mee ing o Expe on Ca bon Dioxide Concen a ion and Rela ed T ace s Measu emen s Techniques, To-
on o, Canada, Sep . 2003, WMO-GAW Repo 161, edi ed by D. Wo hy and L. Huang, pp. 146-151.
Businge , J. A., and A. C. Delany (1990), Chemical senso esolu ion equi ed o measu ing su ace luxes by h ee common
mic ome eo ological echniques, J. A mos. Chem., 10, 399-410.
Businge , J. A., and S. P. Oncley (1990), Flux Measu emen wi h condi ional sampling, J. A mos. Ocean. Tech., 7, 349-352.
Canadell, J. G., H. A. Mooney, D. D. Baldocchi, J. A. Be y, J. R. Ehle inge , C. B. Field, S. T. Gowe , D. Y. Hollinge , J. E.
Hun , R. B. Jackson, S. W. Running, G. R. Sha e , W. S e en, S. E. T umbo e, R. Valen ini, and B. Y. Bond (2000),
Ca bon me abolism o he e es ial biosphe e: A mul i echnique app oach o imp o ed unde s anding, Ecosys ems, 3,
115-130.
Delany, A. C., S. P. Oncley, J. A. Businge , and E. Sie e ing (1991), Adap ing he condi ional sampling concep o a ange
o di e en chemical species, pape p esen ed a Se en h symposium on me eo ological obse a ions and ins umen s,
Ame ican Me eo ological Socie y, Bos on, New O leans, La., 14-18 Janua y 1991.
Desja dins, R. L. (1977), Desc ip ion and e alua ion o a sensible hea lux de ec o , Bounda y-Laye Me eo ol., 11, 147-154.
Ehle inge , J. R., D. R. Bowling, L. B. Flanagan, J. Fessenden, B. Hellike , L. A. Ma inelli, and J. P. Ome o (2002), S able
iso opes and ca bon cycle p ocesses in o es s and g asslands, Plan Biology, 4, 181-189.
Foken, T., R. Dlugi, and G. K amm (1995), On he de e mina ion o d y deposi ion and emission o gaseous compounds a
he biosphe e-a mosphe e in e ace, Me eo ol. Z., 4, 91-118.
Foken, T., M. Göckede, M. Maude , L. Mah , B. Ami o, and W. Munge (2004a), Pos - ield da a qual iy con ol, in Hand-
book o Mic ome eo ology, edi ed by X. Lee, e al., pp. 181-208, Kluwe , Do d ech .
Fo iadi, A. K., F. Lohou, A. D uilhe , D. Se ca, Y. B une , and R. Delmas (2005), Me hodological de elopmen o he condi-
ional sampling me hod. Pa I: Sensi i i y o s a is ical and echnical cha ac e is ics, Bounda y-Laye Me eo ol., 114,
615-640.
Gao, W. (1995), The e ical change o coe icien b, used in he elaxed eddy accumula ion me hod o lux measu emen
abo e and wi hin a o es canopy, A mos. En i on., 29, 2339-2347.
Ge s be ge , P., T. Foken, and K. Kalbi z (2004), The Lehs enbach and S eink euz Ca chmen s in NE Ba a ia, Ge many, in
Biogeochemis y o Fo es ed Ca chmen s in a Changing En i onmen : A Ge man case s udy, edi ed by E. Ma zne , pp.
15-41, Sp inge , Be lin.
Guen he , A., W. Baugh, K. Da is, G. Hamp on, P. Ha ley, L. Klinge , L. Vie ling, P. Zimme man, E. Allwine, S. Dil s, B.
Lamb, H. Wes be g, D. Baldocchi, G. Ge on, and T. Pie ce (1996), Isop ene luxes measu ed by enclosu e, elaxed eddy
accumula ion, su ace laye g adien , mixed laye g adien , and mixed laye mass balance echniques, J. Geophys. Res.,
101, (D13), 18555–18568.
Hübne , J (2010), Einsa z o be ei ung eine REA-Anlage übe Wiesen lächen, Diploma Thesis, Uni e si y o Bay eu h,
Ge many, h p://www.baycee .uni-bay eu h.de/mm/de/ op/diss/88759/DA_Joe g_Huebne .pd
Jo dan, A., and W. A. B and (2003), Technical Repo : MPI-BGC, Ge many, in Repo o he Ele en h WMO/IAEA Mee ing
o Expe s on Ca bon Dioxide Concen a ion and Rela ed T ace Measu emen Techniques, Tokyo, Japan, Sep . 2001,
WMO-GAW Repo 148, edi ed by S. To u and S. Kazu o, pp. 149-153.
Kaplan, J. O., I. C. P en ice, and N. Buchmann (2002), The s able ca bon iso ope composi ion o he e es ial biosphe e:
Modeling a scales om he lea o he globe, Glob. Biogeochem. Cycles, 16, 1060, doi:10.1029/2001GB001403.