Analysis of the temperate tree growth response to climate change through a modelling approach

Transcription

1 Analysis of the temperate tree growth response to climate change through a modelling approach Characterization of prediction uncertainties PhD STUDENT: LOUIS DE WERGIFOSSE SUPERVISOR : MATHIEU JONARD (UCL ELIE) CO-SUPERVISOR: HUGUES GOOSSE (UCL ELIC)

2 Methods to evaluate climate change impact on forests Long-term monitoring of forest ecosystems Observational studies of species distribution according to climate Environment modification experiments Process-based modelling 2

3 Methods to evaluate climate change impact on forests Long-term monitoring of forest ecosystems Observational studies of species distribution according to climate Environment modification experiments EVALUATION Process-based modelling KNOWLEDGE INTEGRATION 3

4 How to model the climate change impact on forests? Complex process-based models at the stand scale 4 (Dufrêne et al, 2005)

5 Current approach limitations Complex process-based models Hard to calibrate (uncertainties) and to validate at the stand scale Management poorly or not accounted for Unadapted to irregular and mixed stands Hard to compare with retrospective individual meaurements 5

6 Toward a better inclusion of the management impact Hybrid (observation and process-based), spatially explicit model at the tree level Climate sensitivity unsatisfayingly considered APAR GPP NPP 6

7 Research objectives Simulate tree growth dynamics according to different climate scenarios and forestry practices in different areas (Wallonia, France, Europe) Develop a methodology to evaluate climate change impact while characterizing the uncertainties Better understanding of the temperate tree growth response to climate change 7

8 Research project description IMPROVEMENT EVALUATION CLIMATE SCENARIOS Tree-level growth model HETEROFOR PREDICTIONS BAYESIAN CALIBRATION 8

9 1 st step Inclusion of climate sensitivity in HETEROFOR and evaluation IMPROVEMENT EVALUATION Tree-level growth model HETEROFOR 9

10 1 st step Inclusion of climate sensitivity in HETEROFOR and evaluation APAR GPP NPP Photosynthesis PAR used efficency Farquhar equations Stomatal regulation Inclusion of a «Water balance «Bilan» hydrique module» Temperature-dependant respiration Inclusion of a phenological module 10

11 Photosynthesis module development Use of Castanea photosynthesis module developped on CAPSIS platform by Dufrene, Davi, François, Le Maire, Le Dantec et al during a first time. (Cfr. Dufrene et al, 2005 for more informations) 11

12 Water balance module development 12

13 Water balance module development CC ffffffffffffff = ssss Area llllllll_ ssss. cc ffffffffffffff_ ssss _ mmmmmm + 1 WWWW WWWW mmmmmm or. cc ffffffffffffff_ ssss _ mmmmmm cc ffffffffffffff _ ssss _ mmmmmm CC ffffffffffffff = (AAAAAAAA llllllll_ssss. cc llllllll_ssss ) sspp 13

14 Water balance module development Variables calculated using empirical equation from André et al (2008): ssttttttffllllll = aa + bb. CCCCC + cc. RRRRRRRR + dd. CCCCC. RRRRRRRR giving CC bbbbbbbb = (aa + bb CCCCC) %sssseeeeeeeeeeee = (cc + dd CCCCC) RRRRRRRR AArrrrrr ssssssssss RRRRiiii %tttttttttttttttttttt = 1 %ssssssmmmmmmmmmm 14

15 15 Water balance module development Evapotranspiration calculated via Penman-Monteith equation (1965): RR + ρρ. cc pp. VVVVVV rr λλ. EEEE = aa + γγ rr aa + rr ss rr ss Foliage evaporation (eddy-covariance approach): rr aa = 1 gg aa = rr ss = 0 WWWW ll 1 Bark evaporation: rr ss = 1 gg ss_bbbbbbbb = gg ss_bbbbbbbb_mmmmmm + (gg ss_bbbbbbbb_mmmmmm gg ss_bbbbbbbb_mmmmmm ) ppppppppss bbbbbbbb CC bbbbbbbb 1 Soil (first horizon) evaporation: rr ss = 1 gg ss_ssssssss = gg ss_ssssssss_mmmmmm + (gg ss_ssssssss_mmmmmm gg ss_ssssssss_mmmmmm ) pppppppppppppp ffffffffffff_ffffffffff 1

16 Water balance module development Water uptake by roots 16

17 Water balance module development 17

18 Water balance module development The matric potential Ψ is calculated from pressure head h determined, in turn, by van Genuchten equation (1980): SS = θθ θθ rr θθ ss θθ rr 1 nn nn SS = 1 + (αα h ) nn Parameter values come from Vereecken pedotransfer functions (Weynants et al, 2009), which require information about organic, clay and sand contents and bulk density 18

19 Water balance module development Vertical movements through horizons are regulated by soil conductivity, gravity and water potential gradient. Conductivity is obtained with Mualem - van Genuchten model equation (1980): KK = KK 0 SS λλ nn 1 1 SS 1 1 nn 1 nn 2 For organic horizons, parameter values from Dettmann et al. (2014) are used. They emanate from peat soil observations. For mineral horizons, pedotransfer equations elaborated by Weynants et al. (2009) are used. 19

20 Water balance module development Surplus is currently calculated according to the assumption that once the horizon is saturated, excedent water is transferred to the next horizon. Some improvements are planned. Drainage and capillary rise both follow the same equations: DD = KK hrr,hrr+1 24 δδh mm δδδδ + 1 AA ssssssssss 10 KK hrr,hrr+1 = KK hrr ee hrr + KK hrr+1 ee hrr+1 ee hrr + ee hrr+1 δδh mm δδδδ = h hrr+1 h hrr ee hrr + ee hrr Comparisons between model results and data, however, show that using the minimal conductivity value gives better results than when the conductivity is averaged 20

21 Phenological module development 21

22 Phenological module development Chilling rate (Chuine, 2009): RR cc = 1 1+ee CCCC(TT CCCC)2 +CCCC(TT CCCC), 5 TT 10 0, TT > 10 oooo TT < 5 SS cc = tt tt0 RR cc (TT tt ), tt = tt 1 iiii SS cc > CC 22

23 Phenological module development Forcing rate (Chuine, 2009): RR ff = 1+ee 1 FFFF(TT FFFF), TT > 0 0, TT 0 SS ff = tt tt1 RR ff (TT tt ), tt = BBBB iiii SS ff > FF 23

24 Phenological module development Yellowing rate (Dufrene et al, 2005): RR yyyyyyyy = TT bb_yyyyyyyy TT, TT < TT bb_yyyyyyyy aaaaaa tt tt 3 0, TT TT bb_yyyyyyyy oooo tt < tt 3 SS yyyyyyyy = tt tt3 RR yyyyyyyy (TT tt ), tt = FFFFFFFF iiii SS yyyyyyyy > FF yyyyyyyy 24

25 Model outputs 25

26 Model outputs 26

27 Model outputs 27

28 1 st year Module evaluation with data from a reference site 28

29 2 nd year Selection of PIC forest sites 29

30 3 rd year Climate scenarios constitution and integration of their different variability sources CLIMATE SCENARIOS Tree-level growth model HETEROFOR 30

31 3 rd year Climate scenarios constitution and integration of their different variability sources Future society GHG emissions Global models Dynamic downsc. Statistical downsc. Forest impacts CORDEX RESEARCH 31

32 Downscaling 2 types of method: - Statistical downscaling: Use of statistical correlations between large-scale climate phenomena (values of the atmospheric pressure field) and local climate (monthly averaged temperatures) (Mearns, 2009) - Dynamic downscaling: Use of climate models working at a finer scale and using global climate model results as boundary conditions (Hoar et Nychka, 2008)

33 3-4 th years Inclusion of HETEROFOR uncertainties: bayesian calibration Tree-level growth model HETEROFOR BAYESIAN CALIBRATION 33

34 3-4 th years Inclusion of HETEROFOR uncertainties: bayesian calibration Prior distribution of the parameter values pp DD Θ. pp(θ) pp(dd) = pp Θ DD Posterior distribution of the parameter values Model Experimental data 34

35 Expected results Individual growth comparison climatescenario x social status A1B dominant trees Cumulated radial growth (mm) B1 dominant trees A1B dominated trees B1 dominated trees 35

36 Expected results Individual growth comparison climate scenario x stand type A1B pure stand Cumulated radial growth (mm) B1 pure stand A1B mixed stand B1 mixed stand 36

37 Practical prospects and perspectives Tool for helping forestry practice decisions Sylvicol itineraries enhancing resilience to tackle cl. change induced risks: Choice of resistant species Stand types Forestry policies & land use plans Methodology to analyse climate change impacts while characterizing uncertainties : Reusable for other issues 37