1. Using tree ring databases to evaluate regional climate drivers of productivity variability in ORCHIDEE-FM model
Kun Tan1, Flurin Babst2, Ben Poulter1, Philippe Ciais1, David Frank2, Thomas Launois1, Valentin Bellassen1
1 LSCE, Lab. des Sciences du Climat et de l’Environnement, Gif sur Yvette Cedex, France
2 WSL, Swiss Federal Research Institute, Birmensdorf, Switzerland

2. Model introduction ORCHIDEE-Forest-Management version
- simulate individual growth of trees in an even-aged stand
Specific add-ons to the standard version
age-related decline NPP
age-related limitation of LAI in young stands
age-related allocation ratio between stem and coarse roots, branch mortality and coarse woody debris litter compartment
Forest management module (simulates three main processes)
Annual time step
Distributes the annual stand-level wood increment to individual trees (using modeled or observed increment)
Unmanaged => natural mortality due to self-thinning
Managed => the timing and intensity of thinning or clear-cuts.
Bellassen et al., 2010, Ecological Modelling, doi: /j.ecolmodel

3. Model introduction Advantage of FM version versus standard version
Ability to simulate annual ring width of individual trees, and its variability within a stand
Better representation of tree and stand growth and mortality

4. Tree-ring modeling at 5 eddy-covariance sites
Climate data
CRUNCEP: , 0.5×0.5deg, 6 hourly time step
WCLIMCRU: , 1km×1km, monthly time step
Tree ring measurements at 5 sites cored by WSL

5. Tree-ring modeling at 5 eddy-covariance sites
1. Comparisons between observed and ORCHIDEE-FM simulated TR width

6. Tree-ring modeling at 5 eddy-covariance sites
2. Comparisons between observed and ORCHIDEE-FM simulated woody NPP
Drought 1976
Shortcomings of the model:
Over-estimated NPP for young trees
Underestimates lag effects of droughts

7. Comparing tree-ring series with modeled NPP at 1000 sites over Europe
Conifers:
PCAB - Picea abies
PISY - Pinus sylvestris
ABAL - Abies alba
LADE - Larix decidua
PICE - Pinus cembra
Broadleaves:
FASY - Fagus sylvatica
QURO - Quercus robur
QUPE - Quercus petraea
PFTs:
BoNE - boreal needleleaf evergreen
TeNE - temperate needleleaf evergreen
TeBS - temperate broadleaves summergreen

8. Comparing tree-ring series with modeled NPP at 1000 sites over Europe
Models, forcing data, and soil depth data used for NPP simulations
Forcing data
CRUNCEP: 0.5×0.5deg, 6 hourly time step
WCLIMCRU: 1km×1km, monthly time step
Soil depth: JRC-ESDB data: almost less than 1 m and never over 1.5 m

9. Comparing tree-ring series with modeled NPP at 1000 sites over Europe
ORCHIDEE has a peak NPP-temperature a few months before the obs. tree ring
Modeled NPP is too sensitive to temperature & precipitation/soil moisture
High lagged correlations with previous years climate but not addressed in models
Fig. Percentage of sites per PFTs with significant (p = 0.05) positive (top two) and negative (bottom one) climate correlations from previous April (A’) to current September (S)

10. Comparing tree-ring series with modeled NPP at 1000 sites over Europe
Fig. Mean monthly NPP from models for the PFTs during

11. Comparing tree-ring series with modeled NPP at 1000 sites over Europe
Fig. Spatial distributions of spring (MAM) and summer (JJA) temperature and precipitation anomaly, and tree ring width (TRW) anomaly and modeled annual NPP (mean ORCHIDEE modeled NPP - ORC mean NPP and mean LPJ modeled NPP - LPJ mean NPP) anomaly in 1959

12. Comparing tree-ring series with modeled NPP at 1000 sites over Europe
Fig. Spatial distributions of spring (MAM) and summer (JJA) temperature and precipitation anomaly, and tree ring width (TRW) anomaly and modeled annual NPP (mean ORCHIDEE modeled NPP - ORC mean NPP and mean LPJ modeled NPP - LPJ mean NPP) anomaly in 1976

13. Comparing tree-ring series with modeled NPP at 1000 sites over Europe
Conclusions
Models are overly sensitive, especially to spring temperature, may come from
Phenology
Vcmax sensitivity to temperature in cool regions
Vcmax sensitivity to precipitation in dry regions
Over sensitive to summer water stress (deep roots, access to ground waters)
Models could not address high lagged correlations with previous years climate
Seasonal plots show large difference between ORCHIDEE and LPJ for winter NPP for conifers – different lagged correlations; both ORCHIDEE and LPJ show similar seasonal NPP, but different magnitudes for broadleaves – similar lagged correlations

14. Thanks !