Interannual changes in the carbon budget of Europea n forests : detecting hot-spots periods of variability
Le Maire G1,2*, Delpierre N 3, Jung M4, Ciais P2, Reichstein M4, Viovy N2 and CARBOEUROPE PIs
1- 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 sites
of 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 d modelled fluxes
Introduction
Identifying determinants and critical seasons that driv e interannual variability of NEE, GPP and TER is a crucial matter for our comprehension of the biogeochemical carbon (C) cycle . Rapid response of photosynthesis to meteorology, as well a s more inertial changes in C pools (woody biomass, Soil Organi c Matter) contribute to the interannual variability (IAV) of forests C balance. A thorough analysis of the determinism of IAV must therefore consider the flux variability at short time scales for years covering 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 year which 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 e observations from seven European forest sites for the period o f 1998-2005 (figure1) . Then we evaluated the ability of th e ORCHIDEE model to locate the HSP episodes during the year , and to identify their meteorological causes .
Fortunately, we found that the model was rather realistic, and used ORCHIDEE to analyze regional and temporal emerging patterns in the HSP at European scale .
and simulations (red) . Thick lines are HSP, meaning that these periods meets two criteria : (1) significance of the correlatio n (represented by the horizontal lines) and (2) ratio of one-month to
Measured GPP-HSP occu r
predominantly during the active vegetation period for all considere d sites (figure 2, black lines) . Th e distribution of GPP-HSP thro ughout th e year varies between sites. The oceani c conifero us LOO site HSP-GPP are mostly located during spring, while they are more evenly distributed along th e year for both conife rous DETha an d FIHyy sites . GPP-HSP a re consistentl y located during summer for all Beech fo re sts (SOR, HES and HAI), as well as for the PUE evergreen Quercus ilex site . Overall, GPP-HSP appea r principally located during the most photosynthetically active periods at al l sites, for which an anomaly of limite d time-length can have significant impact s on the whole year GPP sum .
Measured TER-HSP are less consistently distributed than GPP-HSP , as the amplitudes of respiration processes are dampened (i .e . show less marked seasonality) compared to that of photosynthesis, such that several periods of the year are likely to play a signifcant pa rt in the IAV of TER fluxes . For most sites, NEE-HSP were concomitant to GPP-HSP (figure2) . This result constitutes an evidence for the predominant influence of GPP fluxes on the IAV of NEE, an d complement the findings established by Luyssae rt et al . (2007) and Reichstei n et al . (2007) for three pine site and fo r spatial NEE gradients across Europe , respectively.
ORCHIDEE simulations appears quite good at detecting the GPP-HSP, bu t shows less efficient for detecting HSP- TER, due to the dependence of TER on complex p rocesses (figure 2) . Thoug h moderate, the agreement between both measured and ORCHIDEE HSP detection time series was not due to chance (-statistics of ag reement, no t shown). This is encouraging, given th e generic parameterizations of ORCHIDEE and its range of applications (regional, global) .
.
Fib I~ r
14F b Lñ J4 ¡Pep SiP Oì3 Nw Nell
'J ff a
~t!!
t1~4T ,~~I [-
44 ♦ U-1 ] 4 i ` j -
L r ¡/~
tyl 4a / s ~~7
ys
{
SIC Ii
~Ii t`a~
k Ir
4X
I
1 ~
e, ,, ;,BE 7A ,
?
r' rli
,,
M-7
Vt s~ fy~
s, SjT
yet .
t e( ( eI
Isf
S'< ~
+.~ r
ï' r Vie{ , y _ ãr
1? C" f b
~t i_ _
1 i(
S ¡ k
1'1 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 r
eachflux
and month of the year .Within HSP areas, correlations between the flux and th e driver
(Taor SWC, y-axis label) are represented: Green areas (not .sign): monthly flux does not correlate with meteorological monthly driver. Red areas (sign .pos): monthly flux correlation with monthly meteorology is significant and positive . Blue areas (sign .neg): monthly flux correlation with monthly meteorology is significant and negative
Mapping the hot-spots of variability across Europ e
We used continental ORCHIDEE simulation s to draw monthly maps of climate-flu x correlations (figure 3). When considering the continental maps, fluxes hotspots and climat e cor relation appear re gionally coherent as a function of latitude, so that we concentrate d on zonal means (i .e . projections along the latitudinal gradient) for fu rther analysis (figure 4).
GPP annual anomalies appear to be drive n by temperature from May to October i n nort hern Europe . They are driven instea d by SWC from July to October in southern Europe (figure 4) . The latitudinal boundary between Ta and SWC limitation propagate s No rthwards fro m May to July. Overall, it appears defined by a potential radiatio n threshold of 8 .8 TJ per m² per yr, that is about 52°N . This result is similar to the analysis of Reichstein et al . (2007) regarding the analysis 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 R annual anomalies are driven by temperature from September to May (figure 4) . In Mediterranean re gions, the TER hot- spots are sensitive to SWC during spring and summer.
During most of the year, and at any latitud e the model identifies less periods of stron g correlation between NEE and Ta or SWC than for gross fluxes . The response of gross fluxes to climate may compensate each other, leading only to weak spatial corre lation between climate and NEE . no rthern Europe . NEE is negatively corre lated to Ta (warmer , i .e. more uptake) and positively correlated t o SWC . Over southern Europe, the NE E variability of hot-spots is positively cor related to Ta in summer (warmer, i .e . less uptake) i n July and October, and negatively correlate d to 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 ;
ReichsteinM,
PapaleD, Valentini R, et al. (2007) Determinants of terrestrial ecosystem carbon balance inferred from European eddy covariance flux sites. Geophysical Research Letters.
Acknowledgements: