Dynamic vegetation modeling, climate coupling

To model the terrestrial biosphere and its response to climate and _pCO2 changes, it is imperative to correctly represent global carbon stocks and fluxes, before being able to discretize them on a spatial grid compatible with climate models. The terrestrial biosphere is made up of the stock corresponding to continental vegetation (350-550 GtC, compared with 3 GtC in the oceanic biosphere), to which must be added soil organic matter (1500-2400 GtC).

Numerical models must take account of_CO2 losses from the atmosphere through gross primary production, which are globally offset by inverse flows linked to the respiration of autotrophic plants and heterotrophic soil organisms, as well as smaller terms such as natural combustions.

Empirical laws are used to numerically represent biological and physiological processes on macroscopic scales. These include the response of primary productivity to _pCO2, and the effects of temperature and humidity on heterotrophic respiration in soils.

The geographical extent of vegetation is known, notably from remote sensing, but it is also necessary to subdivide the biosphere into different plant functional types (PFT) for each grid cell in the model. PFT-specific parameterizations are used to represent processes affecting carbon and water exchange, as well as elements such as nitrogen.

Dynamic Global Vegetation Models (DGVM) have been used since the 1990s to simulate the temporal and spatial evolution of primary production and carbon stocks in vegetation and soils, and to calculate their impact on atmospheric _pCO2. Data-model comparisons for the seasonal _pCO2 cycle provide first-order validation of carbon exchange. Model tests are carried out on an ad hoc basis for free atmosphere experimentation sites (FACE).

The DGVM models are then run in transient mode, perturbed by the variations observed over the last few decades in climate, notably temperature and precipitation, as well as in atmospheric _pCO2. Simulations from around ten modelling groups were compared and compiled for the period 1990 to 2010. For these two decades, the terrestrial biosphere is a sink with a gross intensity of 2.4 GtC/year. This sink increases slightly over the period due to the fertilization effect of_CO2, particularly in the tropics.

An additional modelling step is to couple climate models with models representing the carbon cycle, in particular terrestrial vegetation (DGVM). Some twenty Earth System Models (ESMs) have been used as part of an international program (CMIP5).

For the last century, these ESM models have been perturbed by variations in natural (e.g. volcanoes, sun) and anthropogenic (greenhouse gases) climate forcings. Simulated climate variations are broadly in line with observations of atmospheric surface temperature. MSEs are also used to calculate carbon fluxes between the atmosphere and the terrestrial biosphere. The gross flux to the terrestrial biosphere has been rising steadily since the 1960s. This gross biospheric sink fluctuated between 2 and 3 GtC/year over the period 1990-2010, with extreme values (3 to 4 or 1 to 0) corresponding to ENSO variability and volcanic eruptions.

It is possible to project into the future by perturbing numerical models with an increase in _pCO2 linked to the continued burning of fossil fuels. For an intensive scenario with a _pCO2 increase to 800 ppm in 2 100 followed by stabilization until 2200, the simulated climate warms by 4 ^°Coverall, and shows a slight increase in continental precipitation, with strong contrasts between geographical areas. In response, vegetation models show wide regional fluctuations in plant types, and significant changes in net ecosystem productivity: increases at mid and high latitudes, and decreases in certain tropical regions such as Amazonia.

To better understand the respective influences of climate and _pCO2, it is possible to artificially separate the two effects in numerical modelling. The impact of_CO2 alone results in generalized fertilization across all zones. The impact of climate alone presents more contrasts, with some localized increases, but above all decreases in productivity in the tropics. The balance of these changes is reflected in the evolution of net primary production, with a strong increase linked to the dominant effect of_CO2, weakly offset by the climatic effect. To complete the carbon balance, it is necessary to take into account variations in soil respiration, which is increasing as a result of climate impact. Nevertheless, the biosphere still behaves like a sink during the simulated period, but its intensity would erode from 2050 onwards.

As part of the CMIP5 program, a dozen ESM models have been used to calculate the impact of the carbon cycle on atmospheric _pCO2, and to assess any feedback on the climate itself. Overall, the feedbacks from climate-carbon coupling lead to an increase in atmospheric _pCO2 compared with the forcing from uncoupled climate simulation. At the end of a century, warming is therefore slightly greater than for the uncoupled simulation, reflecting a systematically positive climate feedback (i.e. amplification of the disturbance).