ORCHIDEE-TRUNK optimization for CMIP7
ROUND1 (11/04/2026)

Select: config PFT2 (15 sites) PFT3 (30 sites) PFT4 (15 sites) PFT5 (15 sites) PFT6 (15 sites) PFT6opt (8 sites) PFT7 (15 sites) PFT8 (15 sites) PFT9 (30 sites) PFT10 (15 sites) PFT10opt (8 sites) PFT11 (15 sites) PFT12 (15 sites) PFT12opt (8 sites) PFT12optb (8 sites) PFT13 (14 sites) PFT13opt (8 sites) PFT13optb (8 sites) PFT13s2 (1 sites) PFT14 (30 sites) PFT15 (15 sites) PFT15opt (8 sites) PFT15optb (8 sites)

OPTIMIZATION

OptA

Done : PFT6opt, others stopped due to no improvement
8 sites per PFT with max spread across globe
7 control variables (fluxlat, GPP, LAI, ng_ratio, lr_ratio, bm_ratio, ge_ratio)
10 optimized parameters (K_LATOSA_MIN, K_LATOSA_FRAC, K_BELOWGROUND, VMAX_UPTAKE_NH4, VMAX_UPTAKE_NO3, COEFF_MAINT_INIT, SLAINIT, NUE_OPT, LONGEVITY_SAP, DECIDUOUS_RESERVE)

OptB

Done : PFTXXopt, where XX=10,12,13,15 (epsilon 1e-4 from prior)
Done : PFTXXoptb, where XX=12,13,15 (epsilon 1e-3 from prior)
8 sites per PFT with max spread across globe
2 control variables (GPP, LAI) + 5 monitor variables (fluxlat, ng_ratio, lr_ratio, bm_ratio, ge_ratio)
12 optimized parameters : 5 allocation (K_LATOSA_MIN, K_LATOSA_FRAC, VMAX_UPTAKE_NH4, VMAX_UPTAKE_NO3, NUE_OPT) and 7 phenology (GDD_THRESHOLD, HUM_MIN_TIME, PHENO_GDD_CRIT_B, PHENO_GDD_CRIT_C, RELSOILMOIST_ALWAYS, SENESCENCE_RATIO, T_ALWAYS)

Update on 04/05/2026

1. Model switched to: ORCHIDEE_trunk revision 9506.
2. Configuration: all default settings + TEST_ADJUST_RHETERO=n + MOIGDD_ORCHIDEE_2_0=n.
3. Sites: 8 sites selected from Bacour's ensemble with 'good' priors and maximum spread across globe.
4. Control variables:
• fluxlat (closest pixel from FLUXCOM 1/12°, 8-daily interpolated to daily)
• GPP (closest pixel from FLUXCOM 1/12°, 8-daily interpolated to daily)
• LAI (closest pixel from Copernicus GLS 1km, 10-daily interpolated to daily)
• ng_ratio = NPP / GPP (constant mean value taken from previous optimization settings)
• lr_ratio = LEAF_M_c / ROOT_M_c (constant mean value taken from previous optimization settings)
• bm_ratio = (LABILE_M_c + RESERVE_M_c) / (SAP_M_BE_c + SAP_M_AB_c + HEART_M_BE_c + HEART_M_AB_c) (constant mean value taken from previous optimization settings)
• ge_ratio = GPP / EVAP (constant mean value taken from previous optimization settings)
5. Optimization method: L-BFGS-B.
6. Optimized parameters: K_LATOSA_MIN, K_LATOSA_FRAC, K_BELOWGROUND, VMAX_UPTAKE_NH4, VMAX_UPTAKE_NO3, COEFF_MAINT_INIT, SLAINIT, NUE_OPT, LONGEVITY_SAP, DECIDUOUS_RESERVE.
7. Pre-spinup at each iteration: 50Y for high vegetation PFTs (2-9) and 25Y for low vegetation PFTs (10-15)

Configuration

1. Model version: ORCHIDEE_trunk revision 9469, all default settings.
2. Sites: 254 sites across all PFTs selected by C.Bacour.
3. Forcing: global CRUJRA v2.4 0.5°, closest pixel to each site location.
4. PFT composition: as prescribed by C.Bacour (shown on maps with site location).
5. Spinup: 340Y recycling 1901-1920 forcing with fixed CO2 to 1700 level (277.57 ppm).
6. Transient: 200Y across 1701-1900 recycling 1901-1920 forcing with historical CO2 and Nitrogen.
7. Historical: 1901-2015
8. Used simulation data: 2011-2015 with daily history output level (fluxlat, GPP, LAI).
9. Post processing of simulation data: 15-days running averaging window.
10. Observations:
    • fluxlat from FLUXCOM 1/12°, 8-daily, closest pixel to each site location, interpolated to daily
    • GPP from FLUXCOM 1/12°, 8-daily, closest pixel to each site location, interpolated to daily
    • LAI from Copernicus GLS 1km, 10-daily, closest pixel to each site location, interpolated to daily

Remarks

• Simulations are done with default settings as defined in FG_CRUJRA_SPIN:
• CALCULATE_GPP_PREIND = y
• HACK_KB_M1 = y
• OK_EXPLICITSNOW = y
• OK_FREEZE = y
• OK_READ_FM_MAP = y
• ROUGH_DYN = y
• SOILTYPE_CLASSIF = usda
• STOMATE_READ_CN = n
• TEST_ALB_RT_BGSNOWBARE = y
• USE_RATIO_Z0M_Z0H = y
• USE_SOILC_INSULATION = y
• The data from CRUJRA, FLUXCOM, Copernicus are all taken from the grid cell closest to simulated pixel location. For example, for the site P2S1 we have:
  • Cédric's data : lat=-10.96875, lon=-72.281174
  • CRUJRA data : lat=-10.75, lon=-72.25
  • FLUXCOM data : lat=-10.9583336495, lon=-72.2916670975
  • Copernicus data : lat=-10.9732142856936, lon=-72.2767857141388

Associated google-doc

https://docs.google.com/document/d/1cHxALhRF-fnwMFbZiRGCe7bvwLFiN4Y43Me1U5H7RsY