Fig 1.
Strategy for modeling adaptive management in ORCHIDEE-GM.
See text for symbol definitions.
Fig 2.
Illustration of the simulation protocol, forcing data and initial state for various simulations.
Table 1.
Model results and data used for comparison.
Table 2.
Key model parameters for grassland productivity simulations and their ranges.
Fig 3.
Biological potential productivities simulated by ORCHIDEE-GM of (A) potential net primary productivity (harvested dry matter in forage) in cut grasslands and of (B) potential net primary grazed dry matter used for animal intake in pasture.
Both fields are average values for the period 1995–2004.
Fig 4.
Spatial distribution of (A): potential grassland productivity simulated by ORCHIDEE-GM from cut grasslands (M), (B): actual grassland forage productivity from Smit et al. [20] (S), (C): relative discrepancy between them expressed as log (1+(M-S)/S), and (D): the NUTS administrative units at which statistical data are available.
Simulated and “observed” data from statistics represent both a 10-year average from 1995 to 2004.
Table 3.
Comparison of observed and modeled annual NPP at seven temperate grassland sites.
Fig 5.
Comparison between (A): potential livestock density from ORCHIDEE-GM expressed in livestock unit (LSU) per ha of grassland, under the modeled optimal cut-grazing management scheme, and (B): observed livestock density from FAO [62].
Fig 6.
Spatial distribution of potential mowing frequencies (A) and the length of grazing period (B) simulated by ORCHIDEE-GM.
Both fields are average values for the period 1995–2004.
Table 4.
The contribution of rise in atmospheric CO2, climate change and nitrogen fertilization (including nitrogen deposition) trends on trends in potential productivity.
Fig 7.
Temporal variations of modeled grassland productivity and of livestock density from model simulations and grassland agricultural statistics (see text) respectively.
LSU: livestock unit. All variables are spatial averages over European grasslands masked by CORINE Land Cover map (CLC2000 [56]).
Table 5.
Mean value, standard deviation (SD), and coefficient of variation (CV) of modeled and statistical estimates of grassland productivity and NDVI time series for four European countries, as well as correlation coefficient (r) between each pair.
Fig 8.
Normalized temporal evolution of modeled and observed productivity of grasslands in four European countries between 1973 and 2005.
Observed productivity is derived from Smit et al. [20] based on Eurostat agricultural statistics data (see text).
Table 6.
Ratio of livestock density to grassland productivity in Europe.
Fig 9.
Comparison of modeled (Nmodel) and ‘observed’ (Nobs) grass-fed livestock numbers (‘observed’ numbers are inferred by combining Eurostat agricultural statistics and additional data in Eq 13) at the scale of each NUTS-2 region for the period 1990–2010 using ordinary least squares linear regression forced through the origin.
One point is the average over a NUTS-2 region. A regression coefficient (slope) close to unity with a high coefficient of determination (R2) with P < 0.05 indicates a good model-data agreement. Climate zones follow the Köppen-Geiger classification (see text).
Fig 10.
Comparison of modeled (Nmodel) and ‘observed’ (Nobs) (inferred from Eurostat and other data in Eq 13) grass-fed livestock numbers at the scale of NUTS-0 regions in Europe (countries) for the period 1990–2010.
LSU: livestock unit.