Comparing global models of terrestrial net
primary productivity (NPP): Analysis of the seasonal behaviour of NPP,
LAI, FPAR along climatic gradients across ecotones
A. Bondeau, J. Kaduk, D. W. Kicklighter, participants of "Potsdam ’95" (October 1997)
Spatial and seasonal variations of net primary production (NPP), fraction of absorbed photosynthetically active radiation (FPAR) and leaf area index (LAI) simulated by eleven global biospheric models are analysed using two transects covering a temperature and a precipitation gradient. The temperature transect crosses biomes such as tundra, boreal forest, temperate mixed forest, and temperate deciduous forest in North America. The precipitation transect crosses arid shrublands, savannas, and tropical forests in Africa. Two sites have been chosen from each of the two transects to examine the relationship between seasonal variations in NPP, FPAR and LAI in more detail, through the computation of the monthly absorbed photosynthetically active radiation (APAR) and monthly light use efficiency (LUE). Seasonal variations in the climatic variables drive the seasonality of NPP, and depending if the simulated canopy responds to unfavourable periods or not, the seasonal NPP is determined by the seasonal APAR or the seasonal LUE. For the satellite-driven Production Efficiency Models (PEMs) using a standard climatology, the smooth seasonal variations are generally explained by the satellite observations, but the different strategies for processing the satellite data generate significant variability between models. Canopy Models differ widely, in particular with respect to LAI. This is visible over the evergreen forests, though only a small part of the variability of the NPP seasonal profiles between models is explained by the LAI. Models disagree most on the description of the vegetation structure in savannas, where seasonal NPP is strongly dependent on the description of the canopy through both APAR and LUE. In total, strong and smooth seasonal NPP dynamics are obtained independently of the LAI level. This is because the models use different assumptions about the allocation of carbon and nitrogen to leaves and the structures required to support leaves, about respiration costs, and about the coupling of the carbon, nitrogen and water cycles. Models which are calibrated with biome-representative NPP generate clearly visible biome boundaries on the map of the annual NPP as well as in the seasonal fluxes. Some calibration strategies enhance spatial variability within one biome and thereby generate a strong boundary effect at ecotones.
