Biogeosciences, 11, 6595–6612, 2014
www.biogeosciences.ne /11/6595/2014/
doi:10.5194/bg-11-6595-2014
© Au ho (s) 2014. CC A ibu ion 3.0 License.
Ni ous oxide emission budge s and land-use-d i en ho spo s o
o ganic soils in Eu ope
T. Leppel 1, R. Dechow1, S. Gebbe 1, A. F eibaue 1, A. Lohila2, J. Augus in3, M. D ösle 4, S. Fiedle 5, S. Gla zel6,
H. Höpe 7, J. Jä eoja8, P. E. Læ ke9, M. Maljanen10, Ü. Mande 8, P. Mäki an a11, K. Minkkinen12, P. Ojanen12,
K. Regina13, and M. S ömg en14
1Thünen Ins i u e o Clima e-Sma Ag icul u e, B aunschweig, Ge many
2Finnish Me eo ological Ins i u e, Helsinki, Finland
3Leibniz Cen e o Ag icul u al Landscape Resea ch, Münchebe g, Ge many
4Weihens ephan-T iesdo Uni e si y o Applied Sciences, F eising, Ge many
5Johannes Gu enbe g Uni e si y, Mainz, Ge many
6Uni e si y o Vienna, Vienna, Aus ia
7LBEG S a e Au ho i y o Mining, Ene gy and Geology, Hanno e , Ge many
8Uni e si y o Ta u, Ta u, Es onia
9Aa hus Uni e si y, Depa men o Ag oecology, Tjele, Denma k
10Uni e si y o Eas e n Finland, Kuopio, Finland
11Finnish Fo es Resea ch Ins i u e, Van aa, Finland
12Uni e si y o Helsinki, Depa men o Fo es Sciences, Helsinki, Finland
13MTT Ag i ood Resea ch Finland, Jokioinen, Finland
14Swedish Uni e si y o Ag icul u al Sciences, Uppsala, Sweden
Co espondence o: T. Leppel ([email p o ec ed]und.de)
Recei ed: 7 May 2014 – Published in Biogeosciences Discuss.: 16 June 2014
Re ised: 15 Oc obe 2014 – Accep ed: 19 Oc obe 2014 – Published: 1 Decembe 2014
Abs ac . O ganic soils a e a main sou ce o di ec emis-
sions o ni ous oxide (N2O), an impo an g eenhouse gas
(GHG). Obse ed N2O emissions om o ganic soils a e
highly a iable in space and ime, which causes high unce -
ain ies in na ional emission in en o ies. Those unce ain ies
could be educed when ela ing he upscaling p ocess o a
p io i-iden i ied key d i e s by using a ailable N2O obse -
a ions om plo scale in empi ical app oaches. We used
he empi ical uzzy modelling app oach MODE o iden i y
main d i e s o N2O and u ilize hem o p edic he spa ial
emission pa e n o Eu opean o ganic soils. We conduc ed a
me a-s udy wi h a o al amoun o 659 annual N2O measu e-
men s, which was used o de i e sepa a e models o di e -
en land use ypes. We applied ou models o a ailable, spa-
ially explici inpu d i e maps o upscale N2O emissions
a Eu opean le el and compa ed he in en o y wi h ecen ly
published IPCC emission ac o s. The inal s a is ical mod-
els explained up o 60% o he N2O a iance. Ou s udy e-
sul s showed ha c opland and g asslands emi ed he highes
N2O luxes 0.98±1.08 and 0.58±1.03gN2O-Nm−2a−1,
espec i ely. High luxes om c opland si es we e mainly
con olled by low soil pH alue and deep-d ained g ound-
wa e ables. G assland ho spo emissions we e s ongly e-
la ed o high amoun o N- e ilize inpu s and wa me win e
empe a u es. In con as , N2O luxes om na u al pea lands
we e p edominan ly low (0.07±0.27gN2O-Nm−2a−1) and
we ound no ela ionship wi h he es ed d i e s. The o-
al in en o y o di ec N2O emissions om o ganic soils in
Eu ope amoun up o 149.5GgN2O-Na−1, which also in-
cluded luxes om o es and pea ex ac ion si es and ex-
ceeds he in en o y calcula ed by IPCC emission ac o s o
87.4GgN2O-Na−1. N2O emissions om o ganic soils ep-
esen up o 13% o o al Eu opean N2O emissions epo ed
in he Eu opean Union (EU) g eenhouse gas in en o y o
2011 om only 7% o he EU a ea. The eby he model
demons a ed ha he majo pa (85%) o he in en o y
Published by Cope nicus Publica ions on behal o he Eu opean Geosciences Union.
6596 T. Leppel e al.: N2O ho spo s om EU o ganic soils
is induced by an h opogenic managemen , which shows he
signi ican educ ion po en ial by ewe ing and ex ensi ica-
ion o ag icul u ally used pea soils.
1 In oduc ion
Ni ous oxide (N2O) is a na u al ace gas wi h inc easing
abundance in he a mosphe e and adia ion-en o cing p ope -
ies. Soil p ocesses a e he dominan sou ce o e es ial N2O
and con ibu e abou 70% o he o al ne emission budge o
N2O (Mosie , 1998). Maljanen e al. (2010) showed ha N2O
emissions om o ganic soils in No dic coun ies a e 4 imes
highe in compa ison o luxes om mine al soils. In Eu ope
abou 7% o he land a ea is co e ed by o ganic soils, o -
en also called pea soils, acco ding o Mon ana ella e al.
(2006). The N2O luxes om na u al, wa e logged o ganic
soils a e low. D ainage and cul i a ion lead o N mine al-
iza ion om deg ading pea , and consequen ly N2O p oduc-
ion (Wild e al., 1998; Regina e al., 2004) ia ni i ica ion
and deni i ica ion p ocesses (Fi es one and Da idson, 1989).
Thus a , la ge-scale es ima es ha e been based on s a ic
emission ac o app oaches, which only pa ly e lec land
use, clima e, soil nu ien o d ainage s a us. A egional s udy
om Es onia ound signi ican land use di e ences in N2O
emissions om d ained o ganic soils (Mande e al., 2010).
The IPCC ecen ly published new emission ac o s o di -
e en land use ypes, clima e egions and basic soil nu ien
and d ainage ca ego ies o global applica ion in he IPCC
supplemen o na ional g eenhouse gas (GHG) in en o ies
on we lands (IPCC, 2013). Applica ion o emission ac o s in
GHG in en o ies can lead o high unce ain ies (Poulio e al.,
2012). A p esen , he e a e no p ocess-based models o N2O
luxes o o ganic soils ha could be upscaled o explain he
a iabili y o measu ed N2O luxes om Eu opean pea lands
be e han a e age emission ac o s. A success ul upscaling
o an empi ical model could educe he unce ain y o emis-
sion budge s by including unc ional ela ionships o d i ing
pa ame e s. Klemed sson e al. (2005) sugges ed o model
N2O emissions om pea land o es in Sweden wi h an em-
pi ical ela ionship o C/N a io o opsoil, based on obse -
a ions om 12 si es. In G ea B i ain, N2O emissions om
ag icul u al o ganic soils we e modelled wi h a eg ession
o N inpu , wa e - illed po e space (WFPS), soil empe a u e
and land use (Sozanska e al., 2002), based on obse a ions
om 59 si es p edominan ly om mine al soils. The long e -
e ence lis s in he 2013 IPCC supplemen sugges ha he e
a e a la ge amoun o N2O obse a ions in he li e a u e ha
ha e no ye been used o model calib a ion and alida ion.
While some egion- and land-use-speci ic empi ical ela ion-
ships ha e been published (Klemed sson e al., 2005; Mande
e al., 2010), a gene ic unc ional ela ionship be ween N2O
and en i onmen al and managemen d i e s ac oss land use
ca ego ies is missing. This hampe s he de elopmen o man-
agemen s a egies a local, na ional and Eu opean scale o
o ganic soils ha educe an h opogenic N2O emissions. This
s udy aims o
1. de elop gene ic empi ical ela ionships be ween human
and na u al d i e s o N2O applicable ac oss land use
ypes by means o mul i-si e calib a ion wi h all obse -
a ions published un il mid-2013 in Eu ope;
2. de e mine he N2O budge o o ganic soils in Eu ope
and i s a ious sou ces o unce ain y (model, spa ial
d i e da a);
3. de e mine spa ial ho spo s o N2O emissions d i en by
land use, o he human o na u al d i e s and p io i ies
o u u e obse a ions in high-N2O- isk zones;
4. es whe he he new IPCC emission ac o s a e spa ially
ep esen a i e o Eu ope and quan i y po en ial bias.
2 Ma e ial and me hods
2.1 Da abase
The N2O lux syn hesis is based on a me a-s udy o di ec
N2O emissions om o ganic soils. This li e a u e su ey con-
ains N2O obse a ions in Eu ope published un il mid-2013.
All inco po a ed in si u lux measu emen s udies used he
same gas measu emen me hod – he well-es ablished closed-
chambe echnique (Hu chinson and Mosie , 1981). Annual
N2O luxes we e di ec ly aken ou o he publica ions and
all luxes ha ul il he minimum c i e ia o 12 measu e-
men s pe yea we e included in ou analysis. The da abase
con ains he o al amoun o 659 annual lux measu emen s
made on 109 si es in empe a e and bo eal egions in Eu-
ope, sp ead ac oss he main o ganic soil egions (Fig. 1). Nu-
me ous measu emen s came om cen al Eu ope (Ge many,
Ne he lands) and om no he n Eu opean coun ies like Fin-
land, Sweden and Es onia, whe eas he B i ish Isles and eas -
e n and sou he n Eu ope a e unde - ep esen ed in he da ase .
The numbe o measu emen s pe si e di e s om a mini-
mum o 1 annual lux pe iod up o a o al amoun o 59 annual
luxes. Mos o he si es include lux measu emen s om di -
e en plo s ha a y in managemen and en i onmen al con-
di ions. In pa , he expe imen al design was pu posely cho-
sen o dis inguish be ween ea men s o in luences om di -
e en sou ces, e.g. ni ogen e ilize (Vel ho and Oenema,
1995) o wa e con en o opsoil ( an Beek e al., 2010).
We ex ac ed di e se en i onmen al and managemen pa am-
e e s o de i e a wide se o pa ame e s ha can be es ed o
po en ial ela ionships o N2O luxes. The mos equen pa-
ame e s a e lis ed in Table 1 wi h uni s, pa ame e anges
and ac ion o co e age in he s udies. Missing alues o
clima e pa ame e s we e gap- illed wi h da a om he Eu-
opean Clima e Assessmen and Da ase (ECAD), desc ibed
Biogeosciences, 11, 6595–6612, 2014 www.biogeosciences.ne /11/6595/2014/
T. Leppel e al.: N2O ho spo s om EU o ganic soils 6597
Table 1. Lis o po en ial d i ing pa ame e s o N2O wi h uni s, alue mean/ ange and ac ion o measu emen s udies ha co e each
pa ame e . Soil pa ame e s a e ela ed o opsoil laye o 100cm dep h and all pa ame e s a e calcula ed as annual a e age alues, wi h he
excep ion o p ecipi a ion and ni ogen e iliza ion, which a e calcula ed as annual sums.
Name Desc ip ion Uni Mean Min Max F ac ion
(%)
bd Bulk densi y g×cm−30.34 0.03 1.36 69.2
co g O ganic ca bon con en % 36.11 6.7 57.5 79.8
n o To al ni ogen con en % 1.82 0.3 3.9 71.8
ph pH alue – 5.34 3.3 7.63 61
cn Ra io o ca bon and ni ogen – 21.29 9 78.17 80.4
pd Thickness o pea laye m 1.61 0.2 10.2 38.7
ai Ai empe a u e ◦C 6.22 –0.23 11.2 83.5
soil Soil empe a u e ◦C 8.8 1.94 11.78 19.1
pp P ecipi a ion mm 645.2 0 1840 81.6
w G oundwa e able m 0.32 –0.62 1.36 82.2
w ps Wa e - illed po e space % 76.48 41.25 100 13.7
no3 Ni a e concen a ion kg×ha−132.97 0 211.7 13.1
nh4 Ammonia concen a ion kg×ha−128.4 0.33 241 13.1
nmin Mine al ni ogen concen a ion kg×ha−161.37 2.21 241 14.3
n e O ganic and mine al ni ogen e iliza ion kg×ha−143.77 0 713 80.7
in Haylock e al. (2008). All o he da abase e e ences a e
lis ed in Table 6.
2.2 Model de elopmen , calib a ion and alida ion
Fi s ly, he N2O luxes and po en ial d i e s we e analysed
by means o uni a ia e s a is ics. Fu he mo e we in es i-
ga ed he co ela ions be ween luxes and he co espond-
ing d i ing pa ame e s o unde s and in e ac ions and con-
s ain pa ame e combina ions. The speci ied s a is ical anal-
yses we e ca ied ou wi h he p og amming language R (R
De elopmen Co e Team, 2013). Based on hese esul s we
used an empi ical uzzy logic modelling app oach o p edic
N2O luxes based on main d i ing pa ame e s. This da a-
d i en uzzy logic model has been applied o p edic and
upscale annual N2O luxes o ag icul u al mine al soils in
Ge many. The model pe o mance was supe io o o he em-
pi ical app oaches and explained up o 72% o he a iabil-
i y in he da ase . (Dechow and F eibaue , 2011). Ba dossy
e al. (2003) desc ibe he uzzy-based modelling as a as ,
anspa en and pa ame e -pa simonious al e na i e o o he
app oaches. These echniques a e based on he concep o
uzzy logic, a se heo y ha ex ends he bina y logic o ue
(1) and alse (0). I allows o uzzy se s wi h u h alues in
he ange be ween 0 and 1 (deg ee o ul ilmen ) o be had,
and is he e o e able o handle pa ial u h, unce ain ies o
so-called uzziness. The uzzy se s can be used o classi y
ac o domains no only by cons an c isp se s bu also by
di e en unc ion ypes (e.g. iangula , quad a ic) wi h a i-
able membe ship g ade o e he ac o domain. Fu he mo e
i can be u ilized o di ide ac o spaces in o sub-domains
and calcula e all possible combina ions in uzzy in e e ence
schemes (FISs) using uzzy logic algeb a. These FISs can
be me ged in condi ional ule sys ems o model mul i a ia e
p oblems. The app oach is able o model non-linea ela ion-
ships and o ep esen a p io i knowledge ha limi s pa am-
e e spaces o cons ains di ec ions o ela ionships. Ano he
ad an age o uzzy se s in compa ison o o he decision ee
app oaches is he smoo h ansi ion be ween di e en se s
ha allows o mo e accu a e modelling o con inuous a i-
ables. In his s udy iangula uzzy se s o d i ing pa ame-
e s o annual N2O luxes we e calib a ed using a simula ed
annealing echnique o op imize co esponding esponses o
N2O lux measu emen s. We use a o wa d selec ion algo-
i hm in combina ion wi h a sub-da ase , which consis s o
d i e s ha a e a ailable a Eu opean le el, o de e mine he
bes - i ed and egionalizable pa ame e combina ions. The
Nash–Su cli e e iciency (NSE) was used o model assess-
men :
NSE =1−
n
P
i=1Fi
o−Fi
m2
n
P
i=1Fi
o−¯
Fo2.(1)
The coe icien anges om −∞ o 1, whe e he alue o
1 co esponds o a pe ec ma ch and a alue o 0 indi-
ca es an accu acy compa able o he mean o he obse ed
da a. The esidual a iance o he obse ed luxes Fi
oand
he modelled luxes Fi
mmus be smalle han he da a a i-
ance o he obse ed luxes o indica e ha he model is a
be e p edic o han he mean alue o he obse ed da a
Fo. The NSE coe icien is desc ibed as a good indica o
o model p edic ion pe o mance because i is a combined
measu e o sca e and bias (Nash and Su cli e, 1970). The
www.biogeosciences.ne /11/6595/2014/ Biogeosciences, 11, 6595–6612, 2014
6598 T. Leppel e al.: N2O ho spo s om EU o ganic soils
Figu e 1. O e iew o measu emen si es. Size o poin s indica es
numbe o measu emen s pe si e. Backg ound map displays pea -
land dis ibu ion in Eu ope wi h pea co e pe squa e kilome e
om Mon ana ella e al. (2006).
au oma icallyselec edpa ame e combina ions wi h he high-
es NSE measu es abo e 0 ep esen he bes N2O p edic-
o s acco ding o he pa ame e se and pe o mance indica-
o used. The NSEcali and NSEc e e o he NSE coe icien
o he model calib a ion and he alida ion, espec i ely.
Fu he op imiza ion was pe o med by se ing up model
ensembles (MODE) o inal pa ame e combina ions, using
empi ical boo s apping me hods wi h up o 50 indi idual
models, which educes o e - i ing and achie es be e a -
e aged model p edic ions. We ollowed he p ocedu es de-
sc ibed in Dechow and F eibaue (2011).
We alida ed he model esul s by a k- old c oss alida-
ion by s udy si es (Koha i, 1995). The o iginal da ase was
pa i ioned in o ksubse s by s udy si e. A single subsample
was excluded as a alida ion da ase om he calib a ion p o-
cess. All emaining k-1 si es we e used o model calib a ion
and could be alida ed o he independen alida ion se . This
p ocedu e is epea ed k imes un il each si e is used once as a
alida ion da ase . The s udy si es subsamples include a di -
e en numbe o annual luxes, which can con ain up o 15%
o he luxes om he o al da ase . Hence he unequally sized
subsamples can lead o a e y s ic c oss- alida ion esul
in he case o excluding a si e wi h nume ous measu emen s
and high p opo ion o he o al da ase . The calib a ion was
weigh ed by numbe o measu emen s pe si e o a oid o e -
and unde - ep esen a ion o si es wi h small and high num-
be s o lux measu emen s, espec i ely. We also ha e o ake
in o accoun ha he N2O luxes span o e se e al o de s o
magni ude. Hence we applied a loga i hmic ans o ma ion,
Fl
o=ln(F i
o+0.5), (2)
o linea ize he lux ange o be e op imiza ion pe o -
mance. To gene a e models use ul o upscaling, we consid-
e ed only d i ing pa ame e s ha can be egionalized. The e-
o e good p edic o s o N2O luxes like soil ni a e (NO−
3),
ammonium (NH+
4), mine al N con en o CN a io we e no
included in he inal modelling app oach.
2.3 Regionaliza ion
The egionaliza ion desc ibes he applica ion o ou alida ed
uzzy model on EU-wide a ailable inpu da ase s o de i e
consis en N2O emissions o Eu ope. Spa ially explici up-
scaling o he uzzy model was ealized in a geog aphic in o -
ma ion sys em (GIS). We used he open sou ce GRASS GIS
(Ne ele e al., 2012) o p ocess he model inpu da ase s and
p edic N2O emissions a he EU le el. The e o e we de el-
oped and implemen ed se e al GRASS modules o pe o m
uzzy logic modelling in his GIS amewo k. Addi ionally
we conduc ime se ies analysis o clima e and land use da a
by using he empo al amewo k TGRASS (Gebbe and
Pebesma, 2014). The inpu da a a he EU le el is p edom-
inan ly a ailable in as e cell o ma in Lambe azimu hal
equal a ea (LAEA) p ojec ion, wi h he ines esolu ion o
1km×1km g idded da a. Hence we selec ed he LAEA p o-
jec ion and a esolu ion o 1km×1km as a common uni o
a oid da a loss om ans o ma ion p ocesses and as e cell
esampling. The model was applied o pea land a eas in Eu-
ope which a e based on he o ganic soil dis ibu ion map by
Mon ana ella e al. (2006). This da ase se es as a basis o
all spa ial calcula ions in his s udy. The ollowing egional
da ase s we e used o d i ing pa ame e s:
–Land use dis ibu ion:
–CORINE land co e (CLC) om 2006 (Bü ne and
Kosz a, 2007) di e en ia ed in o c opland, g ass-
land, o es , pea ex ac ion and na u al a eas.
–His o ic Land Dynamics Assessmen (HILDA)
(Fuchs e al., 2013) di e en ia ed in o c opland,
g assland (which also con ains na u al a eas) and
o es si es o he la es a ailable yea : 2010.
–Me eo ology: empe a u e and p ecipi a ion om he
ECAD da ase (Haylock e al., 2008). Based on he
daily esolu ion da ase , we calcula ed he 30-yea
(1982–2012) long- e m annual and seasonal (sp ing,
summe , au umn and win e ) minimal, maximal and
mean empe a u es and p ecipi a ions sums.
–Mean annual wa e able: he e a e no spa ially ex-
plici da a a ailable o Eu ope. Mean annual wa e a-
ble was he e o e ep esen ed by land-use-speci ic e-
quency dis ibu ion unc ions o obse ed wa e able in
Biogeosciences, 11, 6595–6612, 2014 www.biogeosciences.ne /11/6595/2014/
T. Leppel e al.: N2O ho spo s om EU o ganic soils 6599
he da abase. The mean alue o he equency dis ibu-
ions was used o egionaliza ion, while he dis ibu ion
se ed o unce ain y assessmen .
–Soil p ope ies: da ase s om he Eu opean soil po al
and he Join Resea ch Cen e (JRC) (Panagos e al.,
2012).
–Topsoil acidi y (Reu e e al., 2008).
–O ganic ca bon con en o opsoil (Jones e al.,
2005).
–Bulk densi y o opsoil (Tik ak e al., 2002). The
Eu opean soil po al p o ides g idded a e ages,
which mix mine al and o ganic soils. Consequen ly,
bulk densi y nei he adequa ely e lec s o ganic
soils no he dependence o bulk densi y on land
use and pea deg ada ion s a us. As o mean an-
nual wa e able, land-use-speci ic equency dis i-
bu ion unc ions o bulk densi y we e used o e-
gionaliza ion.
–Ni ogen e iliza ion based on Hu chings e al. (2012).
The sum o Eu ope-wide annual N2O emissions ep esen s
he emissions om c opland, g assland, o es , pea ex ac-
ion and na u al si es on o ganic soils. Besides he uzzy
model app oach, land-use-s a i ied emission ac o s can also
be u ilized o p edic annual emission budge s. Emission ac-
o s we e de i ed om he N2O lux syn hesis as mean pe
land use ype and compa ed o he IPCC emission ac o s
om he we land supplemen (IPCC, 2013). We used he
good p ac ice guidance o he IPCC ie 1 app oach o calcu-
la e he Eu opean in en o y o N2O emissions om managed
o ganic soils. The IPCC ie 1 app oach s a i ies land use
classes by d ainage, pea ype and clima e zone. The delin-
ea ion be ween he empe a e and bo eal zone can be de i ed
om he IPCC de ini ion applied o clima e da a. D ainage
and pea ype, howe e , a e no a ailable in a spa ially ex-
plici way. We he e o e applied he de aul o nu ien -poo
condi ions in bo eal o es s, nu ien - ich condi ions in em-
pe a e o es s, and deep d ainage in empe a e g asslands.
Spa ial esolu ion and land use de ini ions p oduce signi i-
can unce ain y in he egionaliza ion o N2O emissions. The
unce ain y in land use classi ica ions was assessed by es ing
he sensi i i y o he Eu opean N2O in en o y o he choice
o he land use map, ep esen ed by he wo Eu ope-wide spa-
ially explici map p oduc s CORINE and HILDA. The gen-
e al land use dis ibu ion on o ganic soils can be sepa a ed
in o he o es y-domina ed bo eal zone, he ag icul u al em-
pe a e zone and he main na u al pea land a eas in he sub-
a c ic no he n pa s o Eu ope. N2O emission ho spo s we e
iden i ied on he map oge he wi h ela ed anges o d i e s
sepa a ely o each land-use-speci ic model. In o de o lo-
ca e N2O emission ho spo s in Eu ope, we compu ed he lux
dis ibu ions by land use ca ego y om he N2O emission
map and de ined he luxes abo e he 90 h quan ile as ho spo
emissions o he pa icula land use ca ego y.
2.4 Unce ain y analysis
N2O emissions can a y la gely in space and ime and he
capabili ies o model hese a ia ion a e es ic ed o he size
o he sample da ase and he da a quali y. The e o e i is im-
po an o p opaga e he unce ain ies du ing he modelling
p ocess in o de o be able o es ima e he o e all accu acy o
he model esul . Fo se e al ecosys ems he con idence in e -
al limi s o IPCC emission ac o s o N2O emissions om
pea soils a e g ea e han he mean alues. The modelling
app oach aims o educe his a iabili y by using explana-
o y pa ame e s o p edic N2O luxes. Unce ain y analy-
sis comp ised unce ain ies in inpu pa ame e s and in he
model. The model unce ain y was calcula ed by means o a
uzzy ule-based unce ain y es ima ion (de ails in Dechow
and F eibaue (2011)). I can be desc ibed as he s anda d
de ia ion σ , which is de i ed om he ule-speci ic no mal-
like unce ain y dis ibu ions in Eq. (3):
σ2
=
n
P
i=1DOFiσ2
i
n
P
i=1DOFi
(3)
whe e DOFiis he deg ee o ul ilmen and σ iis he s an-
da d de ia ion o a no mal-like unce ain y dis ibu ion o
ule i. The ule speci ic unce ain y was es ima ed by using
esul s om he c oss alida ion o e s udy si es as a e e -
ence o calcula e he model unce ain y. The inpu pa ame e
unce ain ies we e es ima ed by Mon e Ca lo simula ion wi h
pa ame e a iabili ies aken om a ailable da abases. The
combina ion o inpu and model unce ain y esul s in he
o e all unce ain y es ima ion, which was applied pixel-wise
o unce ain y analysis a he EU le el. The esul ing map
con ains a e age and s anda d de ia ion alues o a no mal-
like dis ibu ion unc ion o N2O emissions o each as e
cell.
The N2O emission budge is he sum o all as e cell al-
ues ha a e loca ed wi hin a de ined a ea. The co espond-
ing unce ain y o he in en o y can be calcula ed by e o
p opaga ion. Spa ially explici modelling in oduces au oco -
ela ion in o he calcula ion o GHG emission in en o ies
and hei unce ain y es ima ion. Wi hou conside a ion o
he spa ial co a iance we would unde es ima e he eal un-
ce ain y. This is a me hodical p oblem ha we sol ed by
conside ing he co a iance in he e o p opaga ion equa ion
o imp o e he unce ain y es ima ion:
σb=
u
u
n
X
i=1
σ2
i+2
n−1
X
i=1
n
X
j=i+1
co ij ,(4)
whe e σi,j is he s anda d de ia ion o a as e cell, indexed
by i, and co ij is he co esponding co a iance be ween all
www.biogeosciences.ne /11/6595/2014/ Biogeosciences, 11, 6595–6612, 2014

6600 T. Leppel e al.: N2O ho spo s om EU o ganic soils
Figu e 2. Box plo s o N2O luxes (a) and mean annual g oundwa-
e able (b) o i e di e en land use ca ego ies (c opland, g ass-
land, pea ex ac ion, o es and na u al si es). N2O luxes a e shown
wi hou ou lie s and nindica es he numbe o measu emen pe ca -
ego y.
as e cell alues, indexed by iand j. We app oxima ed he
co a iances be ween as e cells as a unc ion o dis ance
and calcula ed he co esponding co a iance ma ix o apply
Eq. (4) o he as e map.
3 Resul s and discussion
3.1 S a is ical analysis
The N2O luxes we e log-no mal-dis ibu ed wi h p edom-
inan ly mino luxes be ween −0.1 and 0.1 and ew high
peaks up o 8.11gN2O-Nm−2a−1 om g asslands in he
Ne he lands (Koops e al., 1997). We ound signi ican di -
e ences in lux da a be ween land use ca ego ies; hese a e
shown in Fig. 2. In gene al he highes luxes occu ed on
c opland and g assland si es, whe eas na u al and ewe ed
o ganic soils ea u e low emissions on a e age. Fluxes om
o es si es we e on a e age lowe han he emissions om
c opland and g asslands, bu included some high ou lie s o
up o 6.06gN2O-Nm−2a−1 om Slo enia (Dane ˇ
ciˇ
c e al.,
2010). The pea ex ac ions si es we e only ep esen ed by
35 annual lux measu emen s, which indica ed an a e age
lux o 0.47gN2O-Nm−2a−1 o ac i e and abandoned ex-
ac ion si es. Table 2 lis s he co ela ion coe icien s o
N2O luxes and main d i ing pa ame e s. The mean annual
g oundwa e ables o di e en land use ca ego ies we e
co ela ed o N2O luxes wi h a co ela ion coe icien o
=0.32 (p < 0.05). In addi ion, Fig. 3a shows ha high
N2O luxes occu ed in he ange o mean g oundwa e a-
ble o 0.2 o 0.9m below he su ace. The g oundwa e able
has been ound o be a d i ing pa ame e o N2O in se -
Figu e 3. The sca e plo s show (a) he N2O lux ela ionship o
mean annual g oundwa e able, (b) he ela ionship be ween N e -
iliza ion and N2O luxes o c op- and g assland wi h signi ican
(P < 0.001) linea ela ionship o g assland ( 2=0.26), (c) he
N2O luxes plo ed agains he C/N a ios, and (d) pH alues in e-
la ion o hese C/N a ios including he i ed non-linea unc ion
(ph=15cn−0.36) ( 2=0.5).
e al o he s udies (Ma ikainen e al., 1993; Regina e al.,
1996; an Beek e al., 2010). D ainage inc eases emissions
o N2O, in pa icula o nu ien - ich o ganic soils and e -
ilized and g azed g assland. The seasonal luc ua ions o
he wa e able could explain mo e a iabili y o N2O emis-
sions, bu his in o ma ion was only a ailable o a small
ac ion o he da ase . The e o e we we e es ic ed o he
use o only he mean annual wa e able in ou analysis. The
N- e iliza ion amoun was also co ela ed wi h N2O luxes
( =0.43, p < 0.05). Figu e 3b sugges s ha his ela ion-
ship is especially s ong o emissions om g asslands. The
N2O luxes plo ed agains he C/N a io indica ed a a io
h eshold a app oxima ely 30–35 below which high luxes
occu in he da ase (see Fig. 3c). This esul p o ides e i-
dence and suppo s he indings o Klemed sson e al. (2005)
ha he C/N a io can be a s ong p edic o o N2O emis-
sions om o ganic soils. Pea mine aliza ion eleases ca bon
as CO2, while ni ogen p e e en ially emains in he soil. Ni-
ogen e iliza ion has a simila ne e ec , and hus bo h p o-
cesses educe he soil C/N a io. The e o e he C/N a io
can be u ilized as an indica o o soil p ocesses and condi-
ions ha igge N2O emissions. Figu e 3d shows ha low
pH alues we e ela ed o high C/N a ios and ice e sa.
The collec ed si e da a e ealed a non-linea ela ionship be-
ween pH alues and co esponding soil C/N a ios. Due o
una ailable da a o C/N a ios a he Eu opean le el, he soil
Biogeosciences, 11, 6595–6612, 2014 www.biogeosciences.ne /11/6595/2014/
T. Leppel e al.: N2O ho spo s om EU o ganic soils 6601
Table 2. Co ela ion ma ix o N2O luxes and po en ial d i ing pa ame e s o he a ailable da ase om o ganic soils in Eu ope. The
pa ame e names a e desc ibed in Table 1.
N2O bd co g n o ph cn pd ai soil pp w w ps nmin n e
N2O 1.00 0.17 −0.10∗0.07 −0.05 −0.19 −0.14∗0.06 0.07 0.11∗∗ 0.32 −0.30 0.10 0.43
bd 1.00 −0.80 −0.39 0.37 −0.48 −0.32 0.34 −0.08 −0.17 0.46 0.07 −0.04 0.25
co g 1.00 0.38 −0.50 0.59 0.27 −0.32 −0.08 0.15 −0.31 −0.12 −0.12 −0.13∗∗
n o 1.00 0.14∗−0.40 0.34 0.04 0.11 −0.04 −0.21 0.07 0.26∗0.16
ph 1.00 −0.64 −0.31 0.06 0.22∗−0.30 0.29 −0.03 0.29∗∗ 0.19
cn 1.00 0.02 −0.22 −0.18 0.15 −0.20 −0.19 −0.36 −0.18
pd 1.00 0.17∗∗ 0.29∗0.10 −0.39 −0.06 −0.20 −0.22
ai 1.00 0.77 0.02 −0.11∗−0.01 0.15 0.16
soil 1.00 0.44 0.15 −0.26 0.27 0.07
pp 1.00 −0.13∗∗ −0.14 0.24∗0.01
w 1.00 −0.39 0.08 0.17
w ps 1.00 0.10 −0.01
nmin 1.00 0.10
n e 1.00
Le el o signi icance: ∗∗ signi ican a P≤0.01,∗signi ican a P≤0.05.
pH ela ionship o C/N a ios was used as pa ial p oxy o
C/N a io in he egionaliza ion. The e is a gene al end ha
managed o ganic soils wi h low C/N a io occu on e ile,
mine o ophic pea soils wi h highe pH alues while high
C/N a ios a e ound in nu ien -poo omb o ophic pea -
lands. Ne e heless, he wide sca e o pH alues o a gi en
C/N a io indica es mo e complex spa ial pa e ns, and pH
also has an independen di ec in luence on N2O o ma ion
(see below). Se e al o he s udies ound e idence o clima e
in luence a pa icula pea land si es o egions(Dobbie e al.,
1999; Sozanska e al., 2002; Lohila e al., 2010), which can
be con i med in he ollowing land-use-s a i ied models.
3.2 Model calib a ion and alida ion
3.2.1 Comple e da ase
We applied he uzzy logic model app oach o he en i e
lux da ase , which esul s in he bes - i ed model ensemble
(NSEc =0.12) o ou co a ia es (bulk densi y, g ound-
wa e able, mean win e empe a e and annual p ecipi a-
ion). The s ochas ic a iabili y wi hin he da a p e en s he
gene ic model app oach om p edic ing he measu ed luxes
accu a ely. Thus alida ion esul s we e unsa is ac o y and
we in es iga ed u he imp o emen s using da a pa i ion-
ing wi h ca ego ical pa ame e s such as land use ca ego y,
pea ype and clima e zone. The pea - ype-s a i ied da ase ,
sepa a ed in o bog, en and shallow pea soils, esul s in im-
p o ed model i s o each pea ype. Pea ype, howe e , can-
no be egionalized due o he lack o Eu opean spa ially ex-
plici maps. In con as o F eibaue and Kal schmi (2003),
whe e N2O luxes om empe a e and sub-bo eal clima es
on mine al soils showed di e en mean and maximum emis-
sions, we ound no signi ican di e ences be ween clima e
zones o N2O luxes on o ganic soils. Hence he da a pa -
i ioning by clima e zones had no imp o ing e ec on he
model pe o mance. We achie ed he bes model esul s o
land use s a i ica ion and sepa a ely de eloped uzzy logic
models o c opland, g assland, o es and ex ac ion si es.
The e o e each land use model has a di e en numbe and
ange o obse a ions, as well as di e en co a ia es. Table 3
gi es an o e iew o he land-use-speci ic N2O lux da a and
co esponding model pe o mances.
3.2.2 C opland
The bes - i ed c opland model has a model e iciency o
NSE=0.63 and was calib a ed o h ee pa ame e s – opsoil
pH, he mean g oundwa e able and he annual p ecipi a ion
amoun . These model co a ia es we e alida ed o 40 ob-
se a ions om 20 si es on which all h ee model pa ame e
we e a ailable in ou da ase . The ange o N2O luxes om
he c opland sub-da ase (−0.02, 3.70) in gN2O-Nm−2a−1
was compa able o he ange o he comple e c opland da ase
(−0.02, 6.10). Only a ew ex emely high luxes we e ex-
cluded, and so he mean alues a e equi alen . Using his
sub-da ase , we we e able o achie e he bes model i o
NSEc =0.41 in e ms o an independen c oss alida ion
(compa e Fig. 4).
As has been men ioned in Sec . 3.1, he opsoil pH o c op-
lands was no only co ela ed o N2O emissions ( =−0.53,
p < 0.001) bu also signi ican ly o he C/ N a io ( =−0.68,
p < 0.001). Mø k ed e al. (2007) sugges ed he soil pH o
be a s ong con olling ac o o N2O luxes because i a -
ec s he N2O p oduc ion p ocesses o bo h deni i ica ion
and ni i ica ion. Addi ionally hey s a ed ha low-pH soils
ha e highe N2O/N2p oduc ion a ios and hus highe po-
en ial N2O emissions. The desc ibed e ec is also obse -
able o luxes om c oplands on o ganic soils. Weslien e al.
(2009) also ound a s ong nega i e co ela ion o soil pH and
N2O emissions in hei da a. They a gued ha he dini ogen
www.biogeosciences.ne /11/6595/2014/ Biogeosciences, 11, 6595–6612, 2014
6602 T. Leppel e al.: N2O ho spo s om EU o ganic soils
Table 3. Lis o calib a ed and alida ed N2O uzzy logic models
wi h co a ia es ha a e desc ibed in Table 1 (Pa ame e s), numbe
o lux measu emen s (N lux) and model pe o mances o calib a-
ion (NSEcali) and c oss alida ion (NSEc ) o di e en land use
ca ego ies, espec i ely.
Land use Pa ame e s N lux NSEcali NSEc
C op w , ph, pp 40 0.63 0.41
G ass n e , ai win e , 96 0.68 0.58
pp au umn
Fo es w , ph, ai 60 0.66 0.25
Ex ac ion bd, pp, ai win e 21 0.89 0.28
Na u al – 132 – –
oxide educ ase is inhibi ed by acidic pH and can hus en-
hance N2O emissions (Fi es one and Da idson, 1989; Skiba
and Smi h, 1993). This esul is suppo ed by he indings o
Liu e al. (2010), who ound a s ong nega i e co ela ion be-
ween he N2O/N2p oduc a io o deni i ica ion and soil
pH.
The second impo an pa ame e in he model, he g ound-
wa e able, is well known as a p oxy o oxygen a ailabil-
i y in opsoil and can he e o e signi ican ly con ol he N2O
p oduc ion p ocesses (Regina e al., 1996; an Beek e al.,
2010). We ound a co ela ion be ween N2O and g oundwa-
e able in he c opland da ase which con i med his signi -
icance ( =0.31, p < 0.05). The model indica es ha deep
d ainage induces highe luxes o N2O. In con as o Fig. 3a,
which includes all land use ca ego ies, he model s uc u e
o he ela ionship o g oundwa e able and N2O luxes o
c oplands only was linea and no in he o m o a hump-
shaped, non-linea cu e. The sub-da ase o c oplands in-
dica ed a linea inc ease in N2O luxes wi h deep d ainage.
Fu he mo e, p ecipi a ion eme ged as he hi d model com-
ponen . P ecipi a ion inc eases he WFPS in opsoil and can
igge N2O lux peaks immedia ely a e he ain e en s
(Dobbie e al., 1999; Dobbie and Smi h, 2003). High annual
p ecipi a ion amoun s can inc ease he p obabili y o such
N2O peak lux e en s in d ained ag icul u ally used o ganic
soils.
The expec ed ole o N e ilize , i.e. as a N2O emission
ampli ie on c oplands (Vel ho and Oenema, 1995; Skiba
e al., 1998), could no be con i med in ou modelling ap-
p oach. Bo h he s a is ical analysis, shown in Fig. 3b, and he
uzzy modelling app oach ound no signi ican ela ionship
o N2O luxes and N e iliza ion. O ganic soils unde c op-
lands had C/N a ios below 30 and a e likely s ong sou ces
o ni ogen om pea mine aliza ion. Assuming a soil ca bon
loss om mine alized pea o 7.9 Cha−1a−1, as sugges ed
by he IPCC (IPCC, 2013, Table 2.1), i would esul in a
mean N mine aliza ion o app oxima ely 424.7 kgha−1a−1
o c opland si es in ou da abase wi h a e age C/N a ios o
18.6±5.8. This exceeds he maximum amoun o N e ilize
(288.8kgha−1) ha has been applied o c opland si es. The
Figu e 4. Fuzzy model pe o mance o calib a ion and c oss ali-
da ion o N2O luxes om c opland on o ganic soils. The modelled
luxes (xaxis) ep esen he mean lux a es om a model ensemble
o 50 indi idually boo s apped models. The c oss alida ion was
pe o med by excluding one si e pe i e a ion.
Figu e 5. Fuzzy model esul s o calib a ion and c oss alida ion
o N2O luxes om g assland on o ganic soils. The modelled luxes
(xaxis) ep esen he mean lux a es om a model ensemble o
50 indi idually boo s apped models. The c oss alida ion was pe -
o med by excluding one si e pe i e a ion.
es ima ed mean N mine aliza ion sugges s ha , independen
o e ilize applica ion, su icien subs a e o N2O p oduc-
ion is a ailable and ha he N2O p oduc ion is no limi ed by
ex e nal N inpu . All high luxes om c oplands we e mea-
su ed on deeply d ained si es, which is also e lec ed in he
egionaliza ion by using he g oundwa e dis ibu ion wi h
a mean wa e able o 0.58m below su ace. In summa y,
sensi i i y analysis shows ha he c opland model p edic s
he highes emissions on si es wi h a combina ion o deep
d ainage, a soil pH a ound 4.0 and a high amoun o annual
p ecipi a ion, whe eas he lowes emissions occu o soils
wi h highe pH alues and wa e able nea he su ace, e-
ga dless o ain all.
3.2.3 G assland
G asslands a e he bes -obse ed land use ca ego y, ep e-
sen ed by 217 annual lux measu emen s. The au oma ic cali-
b a ion esul s in a uzzy model wi h h ee pa ame e s, which
can explain abou 68% o he a iabili y in he lux da a
(NSE=0.68). The pa ame e s a e ni ogen e ilize amoun ,
Biogeosciences, 11, 6595–6612, 2014 www.biogeosciences.ne /11/6595/2014/
T. Leppel e al.: N2O ho spo s om EU o ganic soils 6603
Figu e 6. Fuzzy model esul s o calib a ion and c oss alida ion
o N2O luxes om o es si es on o ganic soils. The modelled
luxes (xaxis) ep esen he mean lux a es om a model ensemble
o 50 indi idually boo s apped models. The c oss alida ion was
pe o med by excluding one si e pe i e a ion.
mean win e empe a u e and p ecipi a ion in au umn. The
equi ed pa ame e combina ion is a ailable o 96 obse a-
ions om 44 si es ha co e he N2O lux ange o (−0.03,
4.10) wi h a highe mean (¯x=0.67) han he comple e
g assland da ase (¯x=0.58gN2O-Nm−2a−1). The c oss
alida ion could ep oduce nea ly 60% o he a iabili y in
he da a (NSEc =0.58) (Fig. 5). In ag eemen wi h he s a-
is ical analysis (Fig. 3b), we also ound he signi ican ela-
ionship o N2O luxes and N e iliza ion o he g asslands
uzzy model app oach. The amoun o N e iliza ion was
di ec ly co ela ed ( =0.54, p < 0.05) o he luxes om
g assland si es, whe eas no ela ionship was ound o c op-
lands. In ac , he N- e iliza ion amoun was he mos im-
po an model pa ame e . The impo ance o N e iliza ion
has been ecognized in se e al o he s udies on o ganic soils
(Vel ho and Oenema, 1995; Skiba e al., 1998). The di e -
en esponses o g assland and c opland ha e also been ob-
se ed and modelled o N2O luxes om mine al soils (De-
chow and F eibaue , 2011). Fu he mo e di e en sensi i -
i ies o N e iliza ion on empe a e and sub-bo eal ag icul-
u al mine al soils a e discussed in F eibaue and Kal schmi
(2003) and Roeland e al. (2005).
In addi ion o he managemen in luence, he mean win e
ai empe a u e is also co ela ed o N2O luxes ( =0.40,
p < 0.05)and wasiden i ied asasecond impo an modelpa-
ame e . The emissions inc eased wi h ising win e ai em-
pe a u es up o maximum alues app oxima ely a ound 0◦C.
This ela ion o N2O luxes o mean empe a u es in win e
mon hs (Decembe , Janua y and Feb ua y) can be a p oxy
o he amoun o eleased emissions due o eeze– haw cy-
cles as desc ibed in F eibaue and Kal schmi (2003) and
Jungkuns e al. (2006). Al hough he in e ac ion o pa am-
e e s, e.g. ai empe a u e, WFPS and snow co e , ha can
induce eeze– haw cycles is complex and highly a iable,
he model success ully wo ked wi h win e empe a u e as
a simple inpu pa ame e . This is especially use ul wi h e-
ga d o model upscaling a emp s, because he empe a u e,
Figu e 7. Fuzzy model esul s o calib a ion and c oss alida ion
o N2O luxes om pea ex ac ion si es on o ganic soils. The mod-
elled luxes (xaxis) ep esen he mean lux a es om a model
ensemble o 50 indi idually boo s apped models. The c oss alida-
ion was pe o med by excluding one si e pe i e a ion.
as well as he win e empe a u e only, is easily a ailable a
he Eu opean le el.
Au umn p ecipi a ion eme ged as he hi d model com-
ponen . We obse ed a posi i e co ela ion ( =0.50,
p < 0.05) be ween he ain all amoun in au umn mon hs
(Sep embe , Oc obe and No embe ) and he N2O luxes on
g assland si es. As s a ed be o e, p ecipi a ion can inc ease
he WFPS in opsoil and igge N2O luxes (Dobbie e al.,
1999). This s ong s a is ical ela ion be ween au umn p e-
cipi a ion and N2O has no been desc ibed be o e o o -
ganic g asslands, bu ag ees wi h e idence in mine al c op-
lands Dechow and F eibaue (2011). High p ecipi a ion in
au umn lea es we soils in win e , which is a p econdi ion
o eeze– haw peaks o N2O emissions. In summa y, g ass-
lands N2O luxes a e sensi i e o N e iliza ion and sea-
sonal p ecipi a ion and empe a u es. Highes emissions a e
expec ed o in ensi ely managed g asslands wi h high N in-
pu , which a e con olled by win e empe a u e and ain all
e en s in au umn.
3.2.4 Fo es
The measu ed o es N2O luxes in he da ase (n=170)
a e dominan ly loca ed in bo eal (61%) and sub-bo eal e-
gions (22%), whe eas empe a e o es si es make up only
a small pe cen age (17%). These clima ic egions ha e di -
e en mean N2O emissions 0.51, 0.33 and 0.26 ingN2O-
Nm−2a−1 o empe a e, sub-bo eal and bo eal clima es, e-
spec i ely. Howe e he ange wi hin he clima ic egions a e
compa able and no signi ican di e ence be ween mean N2O
luxes is ecognizable. The bes - i ed o es model consis ed
o h ee pa ame e s: mean g oundwa e able, opsoil pH and
he annual mean ai empe a u e wi h a model e iciency o
NSE=0.66. The co esponding sub-da ase consis ed o 60
obse a ions om 38 si es ha co e he N2O lux ange o
(0.01, 6.06) in gN2O-Nm−2a−1, which is almos iden ical
o he comple e o es da ase . The c oss alida ion le sig-
ni ican a iabili y unexplained (NSEc =0.25). Clea ly, he
www.biogeosciences.ne /11/6595/2014/ Biogeosciences, 11, 6595–6612, 2014
6610 T. Leppel e al.: N2O ho spo s om EU o ganic soils
Bee z, S., Liebe sbach, H., Gla zel, S., Ju asinski, G., Buczko, U.,
and Höpe , H.: E ec s o land use in ensi y on he ull g een-
house gas balance in an A lan ic pea bog, Biogeosciences, 10,
1067–1082, doi:10.5194/bg-10-1067-2013, 2013.
Bell, M. J., Jones, E., Smi h, J., Smi h, P., Yelu ipa i, J., Au-
gus in, J., Juszczak, R., Olejnik, J., and Somme , M.: Simula ion
o soil ni ogen, ni ous oxide emissions and mi iga ion scena -
ios a 3 Eu opean c opland si es using he ECOSSE model, Nu .
Cycl. Ag oecosys., 92, 161–181, doi:10.1007/s10705-011-9479-
4, 2012.
Beye , C. and Höpe , H.: G eenhouse gas emissions om ewe -
ed bog pea ex ac ion si es and a Sphagnum cul i a ion si e in
No hwes Ge many, Biogeosciences Discuss., 11, 4493–4530,
doi:10.5194/bgd-11-4493-2014, 2014.
Bü ne , G. and Kosz a, B.: CLC2006 echnical guidelines, Eu o-
pean En i onmen Agency, Technical Repo , 1, 70 pp., 2007.
Chapuis-La dy, L., W age, N., Me ay, A., Cho e, J.-L., and
Be noux, M.: Soils, a sink o N2O? A e iew, Glob. Change
Biol., 13, 1–17, 2007.
Dane ˇ
ciˇ
c, T., Mandic-Mulec, I., S es, B., S opa , D., and Hacin,
J.: Emissions o CO2, CH4and N2O om Sou he n Eu opean
pea lands, Soil Biol. Biochem., 42, 1437–1446, 2010.
Dechow, R. and F eibaue , A.: Assessmen o Ge man ni ous ox-
ide emissions using empi ical modelling app oaches, Nu . Cycl.
Ag oecosys., 91, 235–254, 2011.
Dobbie, K. E. and Smi h, K. A.: Ni ous oxide emission ac o s o
ag icul u al soils in G ea B i ain: he impac o soil wa e - illed
po e space and o he con olling a iables, Glob. Change Biol.,
9, 204–218, 2003.
Dobbie, K. E., McTagga , I. P., and Smi h, K. A.: Ni ous ox-
ide emissions om in ensi e ag icul u al sys ems: Va ia ions be-
ween c ops and seasons, key d i ing a iables, and mean emis-
sion ac o s, J. Geophys. Res., 104, 26891–26899, 1999.
D ösle , M.: T ace gas exchange and clima ic ele ance o bog
ecosys ems, sou he n Ge many, Ph.D. hesis, 2005.
Eickenscheid , T., Heinichen, J., Augus in, J., F eibaue , A., and
D ösle , M.: Gaseous ni ogen losses and mine al ni ogen ans-
o ma ion along a wa e able g adien in a black alde (Alnus
glu inosa (L.) Gae n.) o es on o ganic soils, Biogeosciences
Discuss., 10, 19071–19107, doi:10.5194/bgd-10-19071-2013,
2013.
Eickenscheid , T., F eibaue , A., Heinichen, J., Augus in, J.,
and D ösle , M.: Sho - e m e ec s o biogas diges a e and
ca le slu y applica ion on g eenhouse gas emissions om
high o ganic ca bon g asslands, Biogeosciences Discuss., 11,
5765–5809, doi:10.5194/bgd-11-5765-2014, 2014.
Eu opean Commission: Annual Eu opean Union g eenhouse gas
in en o y 1990–2011 and in en o y epo 2013, Tech. ep.,
Eu opean Commission, h p://un ccc.in / iles/na ional_ epo s/
annex_i_ghg_in en o ies/na ional_in en o ies_submissions/
applica ion/zip/eua-2013-c -27may.zip, 2013.
Fi es one, M. K. and Da idson, E. A.: Mic obiological basis o NO
and N2O p oduc ion and consump ion in soil, ol. 47, John Wiley
& Sons, 1989.
F eibaue , A. and Kal schmi , M.: Con ols and models o es i-
ma ing di ec ni ous oxide emissions om empe a e and sub-
bo eal ag icul u al mine al soils in Eu ope, Biogeochemis y, 63,
93–115, 2003.
Fuchs, R., He old, M., Ve bu g, P. H., and Cle e s, J.: A high-
esolu ion and ha monized model app oach o econs uc ing
and analysing his o ic land changes in Eu ope, Biogeosciences,
10, 1543–1559, doi:10.5194/bg-10-1543-2013, 2013.
Gebbe , S. and Pebesma, E.: TGRASS: A empo al GIS o ield
based en i onmen al modeling, En i on. Modell. So w., 53,
1–12, 2014.
G ønlund, A., S eis up, T. E., Sø ik, A. K., Rasse, D. P.,
and Klø e, B.: Deg ada ion o cul i a ed pea soils in no h-
e n no way based on ield scale CO2, N2O and CH4emis-
sion measu emen s, A ch. Acke P l. Boden., 52, 149–159,
doi:10.1080/03650340600581968, 2006.
Haylock, M. R., Ho s a, N., Tank, A. M. G. K., Klok, E. J., Jones,
P. D., and New, M.: A Eu opean daily high- esolu ion g idded
da a se o su ace empe a u e and p ecipi a ion o 1950–2006,
J. Geophys. Res., 113, D20119, doi:10.1029/2008JD010201,
2008.
Hu chinson, G. L. and Mosie , A. R.: Imp o ed Soil
Co e Me hod o Field Measu emen o Ni ous
Oxide Fluxes, Soil Sci. Soc. Am. J., 45, 311–316,
doi:10.2136/sssaj1981.03615995004500020017x, 1981.
Hu chings, N. J., Reinds, G. J., Leip, A., Wa enbach, M., Bi-
enkowski, J. F., Dalgaa d, T., D agosi s, U., D oue , J. L., Du-
and, P., Mau y, O., and de V ies, W.: A model o simula -
ing he imelines o ield ope a ions a a Eu opean scale o
use in complex dynamic models, Biogeosciences, 9, 4487–4496,
doi:10.5194/bg-9-4487-2012, 2012.
Hy önen, N., Hu unen, J., Shu pali, N., Ta i, N., Repo, M.,
and Ma ikainen, P.: Fluxes o ni ous oxide and me hane
on an abandoned pea ex ac ion si e: e ec o eed ca-
na y g ass cul i a ion, Bio esou ce Technol., 100, 4723–4730,
doi:10.1016/j.bio ech.2009.04.043, 2009.
IPCC: Supplemen o he 2006 IPCC Guidelines o Na ional
G eenhouse Gas In en o ies: We lands, Tech. ep., he na ional
g eenhouse gas in en o ies p og amme, 2013.
Jones, R. J. A., Hiede e , R., Rusco, E., and Mon ana ella, L.: Es i-
ma ing o ganic ca bon in he soils o Eu ope o policy suppo ,
Eu op. J. Soil Sci., 56, 655–671, 2005.
Jungkuns , H. F. and Fiedle , S.: Geomo phology-key egula o o
ne me hane and ni ous oxide luxes om he pedosphe e, Z.
Geomo phol., 49, 429–543, 2005.
Jungkuns , H. F., F eibaue , A., Neu eld , H., and Ba e h, G.: Ni-
ous oxide emissions om ag icul u al land use in Ge many – a
syn hesis o a ailable annual ield da a, J. Plan Nu i . Soil Sci.,
169, 341–351, 2006.
Klemed sson, L., Klemed sson, A. K., Moldan, F., and Weslien, P.:
Ni ous oxide emission om Swedish o es soils in ela ion o
liming and simula ed inc eased N-deposi ion, Biol. Fe il. Soils,
25, 290–295, 1997.
Klemed sson, A. K., Weslien, P., and Klemed sson, L.: Me hane and
ni ous oxide luxes om a a med Swedish His osol, Eu . J. Soil
Sci., 60, 321–331, doi:10.1111/j.1365-2389.2009.01124.x, 2009.
Klemed sson, L., Von A nold, K., Weslien, P., and Gunde sen, P.:
Soil CN a io as a scala pa ame e o p edic ni ous oxide emis-
sions, Glob. Change Biol., 11, 1142–1147, 2005.
Klø e, B., S eis up, T. E., and Hauge, A.: Leaching o nu i-
en s and emission o g eenhouse gases om pea land cul i-
a ion a Bodin, No he n No way, Geode ma, 154, 219–232,
doi:10.1016/j.geode ma.2009.08.022, 2010.
Biogeosciences, 11, 6595–6612, 2014 www.biogeosciences.ne /11/6595/2014/

T. Leppel e al.: N2O ho spo s om EU o ganic soils 6611
Koha i, R.: A s udy o c oss- alida ion and boo s ap o accu acy
es ima ion and model selec ion, in: In e na ional join Con e -
ence on a i icial in elligence, 14, 1137–1145, 1995.
Koops, J., an Beusichem, M., and Oenema, O.: Ni ogen loss om
g assland on pea soils h ough ni ous oxide p oduc ion, Plan
Soil, 188, 119–130, 1997.
K oon, P., Vesala, T., and G ace, J.: Flux measu emen s o CH4and
N2O exchanges, Ag . Fo es Me eo ol., 150, 745–747, 2010.
Laine, J., Sil ola, J., Tolonen, K., Alm, J., Nykänen, H.,
Vasande , H., Sallan aus, T., Sa olainen, I., Sinisalo, J., and Ma -
ikainen, P. J.: E ec o wa e -le el d awdown on global clima ic
wa ming: no he n pea lands, Ambio, 25, 179–184, 1996.
Leibe -Sauhei l, K., Fuß, R., Voig , C., and F eibaue , A.: High
CO2 luxes om g assland on his ic Gleysol along soil ca -
bon and d ainage g adien s, Biogeosciences, 11, 749–761,
doi:10.5194/bg-11-749-2014, 2014.
Liu, B., Mø k ed, P. T., F os egå d, Å., and Bakken, L. R.: Deni i-
ica ion gene pools, ansc ip ion and kine ics o NO, N2O and
N2p oduc ion as a ec ed by soil pH, FEMS Mic obiol. Ecol.,
72, 407–417, 2010.
Lohila, A., Au ela, M., Ha akka, J., Pihla ie, M., Minkkinen, K.,
Pen ilä, T., and Lau ila, T.: Responses o N2O luxes o em-
pe a u e, wa e able and N deposi ion in a no he n bo eal en,
Eu op. J. Soil Sci., 61, 651–661, 2010.
Maljanen, M., Liikanen, A., Sil ola, J., and Ma ikainen, P. J.: Ni-
ous oxide emissions om bo eal o ganic soil unde di e en
land-use, Soil Biol. Biochem., 35, 689–700, 2003.
Maljanen, M., Sigu dsson, B. D., Guomundsson, J., Óska sson, H.,
Hu unen, J. T., and Ma ikainen, P. J.: G eenhouse gas balances
o managed pea lands in he No dic coun ies – p esen knowl-
edge and gaps, Biogeosciences, 7, 2711–2738, doi:10.5194/bg-
7-2711-2010, 2010.
Maljanen, M., Shu pali, N., Hy önen, J., Mäki an a, P., A o, L.,
Po ila, H., Laine, J., Li, C., and Ma ikainen, P. J.: A o es a-
ion does no necessa ily educe ni ous oxide emissions om
managed bo eal pea soils, Biogeochemis y, 108, 199–218,
doi:10.1007/s10533-011-9591-1, 2012.
Mande , U., Uuemaa, E., Kull, A., Kanal, A., Maddison, M.,
Soosaa , K., Salm, J.-O., Les a, M., Hansen, R., Kulle , R.,
Ha ding, A., and Augus in, J.: Assessmen o me hane and ni-
ous oxide luxes in u al landscapes, Landsc. U ban Plann., 98,
172–181, 2010.
Mäki an a, P., Hy önen, J., A o, L., Maljanen, M., Pihla ie, M.,
Po ila, H., Shu pali, N. J., Laine, J., Lohila, A., Ma ikainen, P. J.,
and Minkkinen, K.: Soil g eenhouse gas emissions om a -
o es ed o ganic soil c oplands and cu away pea lands, Bo eal
En i on. Res., 12, 159–175, 2007.
Ma ikainen, P. J., Nykanen, H., C ill, P., and Sil ola, J.: E ec o a
lowe ed wa e able on ni ous oxide luxes om no he n pea -
lands, Na u e, 366, 51–53, 1993.
Mon ana ella, L., Jones, R. J. A., and Hiede e , R.: The dis ibu ion
o pea land in Eu ope, Mi es and Pea land, 1, 2006.
Mø k ed, P. T., Dö sch, P., and Bakken, L. R.: The N2O p oduc
a io o ni i ica ion and i s dependence on long- e m changes in
soil pH, Soil Biol. Biochem., 39, 2048–2057, 2007.
Mosie , A. R.: Soil p ocesses and global change, Biol. Fe il. Soils,
27, 221–229, 1998.
Nash, J. E. and Su cli e, J. V.: Ri e low o ecas ing h ough con-
cep ual models – Pa I: A discussion o p inciples, J. Hyd ol.,
10, 282–290, 1970.
Ne ele , M., Bowman, M. H., Landa, M., and Me z, M.: GRASS
GIS: A mul i-pu pose open sou ce GIS, En i on. Modell. So -
wa e, 31, 124–130, 2012.
Nykanen, H., Alm, J., Lang, K., Sil ola, J., and Ma ikainen, P. J.:
Emissions o CH4, N2O and CO2 om a i gin en and a en
d ained o g assland in Finland, J. Biogeog ., 22, 351–357,
1995.
Ojanen, P., Minkkinen, K., Alm, J., and Pen ilä, T.:
Soil–a mosphe e CO2, CH4and N2O luxes in bo eal o es y-
d ained pea lands, Fo es Ecol. Manag., 260, 411–421,
doi:10.1016/j. o eco.2010.04.036, 2010.
Panagos, P., Van Liedeke ke, M., Jones, A., and Mon ana ella, L.:
Eu opean Soil Da a Cen e: Response o Eu opean policy sup-
po and public da a equi emen s, Land Use Policy, 29, 329–338,
2012.
Peichl-B ak, M.: The In luence o Land Managemen on he Fluxes
o G eenhouse Gases in O ganic Soils, Uni e si ä Hohenheim,
2013.
Pe e sen, S. O., Ho mann, C. C., Schä e , C.-M., Bliche -
Ma hiesen, G., Elsgaa d, L., K is ensen, K., La sen, S. E., To p,
S. B., and G e e, M. H.: Annual emissions o CH4and N2O,
and ecosys em espi a ion, om eigh o ganic soils in Wes e n
Denma k managed by ag icul u e, Biogeosciences, 9, 403–422,
doi:10.5194/bg-9-403-2012, 2012.
Poulio , G., Wisne , E., Mobley, D., and Hun , William, J.: Quan-
i ica ion o emission ac o unce ain y, J. Ai Was e Manage.
Assoc., 62, 287–298, 2012.
R De elopmen Co e Team: R: A language and en i onmen o s a-
is ical compu ing, ISBN 3-900051-07-0. R Founda ion o S a-
is ical Compu ing. Vienna, Aus ia, h p://www.R-p ojec .o g,
2013.
Rees, R. M., Augus in, J., Albe i, G., Ball, B. C., Boeckx, P.,
Can a el, A., Cas aldi, S., Chi inda, N., Chojnicki, B.,
Giebels, M., Go don, H., G osz, B., Ho a h, L., Juszczak, R.,
Kasimi Klemed sson, Å., Klemed sson, L., Medine s, S., Ma-
chon,A.,Mapanda, F.,Nyamanga a,J., Olesen,J.E., Reay, D.S.,
Sanchez, L., Sanz Cobena, A., Smi h, K. A., Sowe by, A., Som-
me , M., Soussana, J. F., S enbe g, M., Topp, C. F. E., an Cleem-
pu , O., Vallejo, A., Wa son, C. A., and Wu a, M.: Ni ous oxide
emissions om Eu opean ag icul u e – an analysis o a iabili y
and d i e s o emissions om ield expe imen s, Biogeosciences,
10, 2671–2682, doi:10.5194/bg-10-2671-2013, 2013.
Regina, K., Nykänen, H., Sil ola, J., and Ma ikainen, P. J.: Fluxes
o ni ous oxide om bo eal pea lands as a ec ed by pea land
ype, wa e able le el and ni i ica ion capaci y, Biogeochem-
is y, 35, 401–418, 1996.
Regina, K., Sy äsalo, E., Hannukkala, A., and Esala, M.: Fluxes o
N2O om a med pea soils in Finland, Eu op. J. Soil Sci., 55,
591–599, 2004.
Reu e , H. I., Lado, L. R., Hengl, T., and Mon ana ella, L.:
Con inen al-scale digi al soil mapping using Eu opean soil p o-
ile da a: soil pH, Hambu ge Bei äge zu Physischen Geog a-
phie und Landscha sökologie, 19, 91–102, 2008.
Roeland , C., Van Wesemael, B., and Rounse ell, M.: Es ima ing
annual N2O emissions om ag icul u al soils in empe a e cli-
ma es, Glob. Change Biol., 11, 1701–1711, 2005.
www.biogeosciences.ne /11/6595/2014/ Biogeosciences, 11, 6595–6612, 2014
6612 T. Leppel e al.: N2O ho spo s om EU o ganic soils
Salm, J. O., Maddison, M., Tammik, S., Soosaa , K., T uu, J., and
Mande , U.: Emissions o CO2, CH4and N2O om undis u bed,
d ained and mined pea lands in Es onia, Hyd obiologia, 692,
1–15, 2011.
Skiba, U. and Smi h, K. A.: Ni i ica ion and deni i ica ion as
sou ces o ni ic oxide and ni ous oxide in a sandy loam soil,
Soil Biol. Biochem., 25, 1527–1536, 1993.
Skiba, U., Sheppa d, L., Macdonald, J., and Fowle , D.: Some key
en i onmen al a iables con olling ni ous oxide emissions om
ag icul u al and semi-na u al soils in Sco land, A mos. En i on.,
32, 3311–3320, 1998.
Sozanska, M., Skiba, U., and Me cal e, S.: De eloping an in en-
o y o N2O emissions om B i ish soils, A mos. En i on., 36,
987–998, 2002.
S ömg en, M., F öbe g, M., and Olsson, M.: G eenhouse gas luxes
om ou d ained o es ed pea lands wi h di e en e ili y in
sou he n Sweden, Bo eal En i on. Res., in e iew, 2014.
Tauchni z, N., B umme, R., Be nsdo , S., and Meissne , R.: Ni ous
oxide and me hane luxes o a p is ine slope mi e in he Ge man
Na ional Pa k Ha z Moun ains, Plan Soil, 303, 131–138, 2008.
Tik ak, A., Nie, D. D., Van De Linden, T., and K uijne, R.: Mod-
elling he leaching and d ainage o pes icides in he Ne he lands:
he GeoPEARL model, Ag onomie, 22, 373–387, 2002.
an Beek, C., Pleij e , M., Jacobs, C., Vel ho , G., an G oenigen, J.,
and Kuikman, P.: Emissions o N2O om e ilized and g azed
g assland on o ganic soil in ela ion o g oundwa e le el, Nu .
Cycl. Ag oecosys., 86, 331–340, 2010.
Vel ho , G. and Oenema, O.: Ni ous oxide luxes om g assland in
he Ne he lands: II. E ec s o soil ype, ni ogen e ilize appli-
ca ion and g azing, Eu op. J. Soil Sci., 46, 541–549, 1995.
Vel ho , G. L., B ade , A. B., and Oenema, O.: Seasonal a ia ions
in ni ous oxide losses om managed g asslands in he Ne he -
lands, Plan Soil, 181, 263–274, doi:10.1007/BF00012061, 1996.
Weslien, P., Klemed sson, A. K., Bö jesson, G., and Klemed sson,
L.: S ong pH in luence on N2O and CH4 luxes om o es ed
o ganic soils, Eu op. J. Soil Sci., 60, 311–320, 2009.
Wild, Klemisch, and P adenhaue : Ni ous oxide and me hane
luxes om o ganic soils unde ag icul u e, Eu op. J. Soil Sci.,
49, 327–335, 1998.
Yamulki, S., Ande son, R., Peace, A., and Mo ison, J. I. L.: Soil
CO2CH4and N2O luxes om an a o es ed lowland aised
pea bog in Sco land: implica ions o d ainage and es o a ion,
Biogeosciences, 10, 1051–1065, doi:10.5194/bg-10-1051-2013,
2013.
Biogeosciences, 11, 6595–6612, 2014 www.biogeosciences.ne /11/6595/2014/