Interannual changes in the carbon budget of Europea nforests : detecting hot-spots periods of variability

Le Maire G1,2*, Delpierre N 3, Jung M4 , Ciais P2 , Reichstein M4, Viovy N2 and CARBOEUROPE PIs1- Fonctionnement et pilotage de écosystèmes de plantation (CIRAD), Montpellier, France ; 2-Lab . Sciences du Climat et de l’Environnement, Gif-sur-Yvette, France

3-LESE Université Pa ris-Sud, Orsay, France; 4-MPI für Biogeochemistry, Jena, German y*guerric .le_maire@cirad.fr

Figure 1 . Location of the seven eddy-flux sitesof the study . Triangles= Pine or Spruce forests (ENF )Disks= Beech forests (DBF) ; Square= Quercus ilex (EBF )HYY= Hyytiala, SOR= Soroe, LOO= Loobos, HAI=Hainich, THA=Tharandt, HES= Hesse, PUE= Puéchabon .

HSP of measured an dmodelledfluxes

IntroductionIdentifying determinants and critical seasons that driv einterannual variability of NEE, GPP and TER is a crucial matterfor our comprehension of the biogeochemical carbon (C) cycle .Rapid response of photosynthesis to meteorology, as well a smore inertial changes in C pools (woody biomass, Soil Organi cMatter) contribute to the interannual variability (IAV) of forests Cbalance. A thorough analysis of the determinism of IAV musttherefore consider the flux variability at short time scales for yearscovering different climatic conditions.

We first developed a method to detect the hot spot periods(HSP) in flux time series . We define HSP as periods of the yearwhich contribute most to the IAV of GPP, TER and NEE fluxes .We further identified climatic drivers associated to HSP periods .

This method was first tested on half hourly eddy covarianc eobservations from seven European forest sites for the period o f1998-2005 (figure1) . Then we evaluated the ability of th eORCHIDEE model to locate the HSP episodes during the year ,and to identify their meteorological causes .

Fortunately, we found that the model was rather realistic, andused ORCHIDEE to analyze regional and temporal emergingpatterns in the HSP at European scale .

and simulations (red) . Thick lines are HSP, meaning that theseperiods meets two criteria : (1) significance of the correlation(represented by the horizontal lines) and (2) ratio of one-month to

Measured GPP-HSP occu rpredominantly during the activevegetation period for all considere dsites (figure 2, black lines) . Thedistribution of GPP-HSP thro ughout th eyear varies between sites. The oceanicconiferous LOO site HSP-GPP aremostly located during spring, while theyare more evenly distributed along th eyear for both conife rous DETha an dFIHyy sites . GPP-HSP a re consistentl ylocated during summer for all Beechfo re sts (SOR, HES and HAI), as well asfor the PUE evergreen Quercus ilexsite . Overall, GPP-HSP appea rprincipally located during the mostphotosynthetically active periods at al lsites, for which an anomaly of limite dtime-length can have significant impactson the whole year GPP sum .

Measured TER-HSP are lessconsistently distributed than GPP-HSP ,as the amplitudes of respirationprocesses are dampened (i .e . showless marked seasonality) compared tothat of photosynthesis, such that severalperiods of the year are likely to play asignifcant pa rt in the IAV of TER fluxes .

For most sites, NEE-HSP wereconcomitant to GPP-HSP (figure2) .This result constitutes an evidence forthe predominant influence of GPPfluxes on the IAV of NEE, an dcomplement the findings established byLuyssae rt et al . (2007) and Reichstei net al . (2007) for three pine site and fo rspatial NEE gradients across Europe ,respectively.

ORCHIDEE simulations appears quitegood at detecting the GPP-HSP, bu tshows less efficient for detecting HSP-TER, due to the dependence of TER oncomplex p rocesses (figure 2) . Thoug hmoderate, the agreement between bothmeasured and ORCHIDEE HSPdetection time series was not due tochance (-statistics of ag reement, no tshown). This is encouraging, given th egeneric parameterizations ofORCHIDEE and its range ofapplications (regional, global) .

.Fib

I~r

14F

b

Lñ

J4

¡Pep

SiP

Oì3

Nw

Ne

ll

'Jff

a~

t!!

t1 ~4T ,~~I [- 4

4 ♦ U-1]

j

i `-4¡/~

L r tyl 4a /s

~~7 y s

{

SIC Ii ~ Iit`a~

k

Ir 4XI

1 ~

e, ,, ;,BE 7A ,

?

r'rli

,,

M-7V

t s~

fy~ s,

S

jT.

t

yet

e(

(e

I I

sfS

'< ~+.~ rï' r Vie{ ,y _ ãr1?

C"

f b~t

i_

_1 i(

S

¡

k1'1

I`-I

l_

(~ I

,cl

( .I

l01

L-. .

k-l

C-•

1-•

'-01

111

I;

-1

I

(4(I

. .

cl

(, :I

l

t

I - .nal

Figure 3. Maps of Europe representing HSP areas (colours), not-HSP in grey fo reach flux and month of the year . Within HSP areas, correlations between the flux and th edriver (Ta or SWC, y-axis label) are represented: Green areas (not .sign): monthly flux does notcorrelate with meteorological monthly driver. Red areas (sign .pos): monthly flux correlation withmonthly meteorology is significant and positive . Blue areas (sign .neg): monthly flux correlation withmonthly meteorology is significant and negative

Mapping thehot-spotsofvariability across Europe

We used continental ORCHIDEE simulation sto draw monthly maps of climate-flu xcorrelations (figure 3). When considering thecontinental maps, fluxes hotspots and climat ecor relation appear re gionally coherent as afunction of latitude, so that we concentrate don zonal means (i .e . projections along thelatitudinal gradient) for fu rther analysis(figure 4).

GPP annual anomalies appear to be drive nby temperature from May to October i nnort hern Europe . They are driven instea dby SWC from July to October in southernEurope (figure 4) . The latitudinal boundarybetween Ta and SWC limitation propagatesNo rthwards from May to July. Overall, itappears defined by a potential radiationthreshold of 8 .8 TJ per m² per yr, that is about52°N . This result is similar to the analysis ofReichstein et al . (2007) regarding theanalysis of the drive rs of spatial (inte rs ite )variability in measured GPP and TER .

No rth of 50°N, the hot-spot related TE Rannual anomalies are driven bytemperature from September to May (figure4) . In Mediterranean re gions, the TER hot-spots are sensitive to SWC during springand summer.

During most of the year, and at any latitud ethe model identifies less periods of stron gcorrelation between NEE and Ta or SWCthan for gross fluxes . The response ofgross fluxes to climate may compensate eachother, leading only to weak spatial corre lationbetween climate and NEE . no rthern Europe .NEE is negatively corre lated to Ta (warmer ,i .e. more uptake) and positively correlated toSWC. Over southern Europe, the NE Evariability of hot-spots is positively cor relatedto Ta in summer (warmer, i .e . less uptake) i nJuly and October, and negatively correlate dto SWC (dryer, i .e . less uptake).

References : Luyssaert S, Janssens IA, Sulkava M, et al. (2007) Photosynthesis drives anomalies in net carbon-exchange of pine forests at different latitudes . Global Change Biology ; Reichstein M, Papale D, Valentini R,et al. (2007) Determinants of terrestrial ecosystem carbon balance inferred from European eddy covariance flux sites. Geophysical Research Letters.Acknowledgements: Sites PIs (Timo Vesala, Ebba Dellwik, Eddy Moors, Werner Kutsch, Corinna Rebmann, Christian Bernhofer, Thomas Grünwald, André Granier, Bernard Longdoz, Serge Rambal, Jean-MarcOurcival), field researchers and students involved in the maintenance of the flux sites.