poster

Increases in temperature and water deficit as a result of climate changes have already impaired forest functioning and might trigger tree dieback world-wide in the near future. To assess the future of forested ecosystems, humans rely on mechanistic models that predict changes in trees and forest functioning as a function of meteorological drivers that are provided by global or regional climate models (GCM or RCM). One large uncertainty when forecasting the forest functioning is associated with the coarse spatial resolution of climate scenarii. In this study we assessed how the spatial resolution in climate forcing provided by the RCM WRF impacted the simulated productivity and distribution of Fagus sylvatica over France. We ran the forest model CASTANEA (that simulates fluxes of carbon and water and forest growth) using the output of WRF at different spatial scales (50 km, 20km, 8km) as forcing climate entries over France. We simulated beech productivity as a surrogate of beech persistence for the reference period of WRF (1988-2008). Because climate variables simulated by WRF exhibited large bias compared to surface observations, WRF was first corrected using the SAFRAN database upscaled at the WRF resolution (50km and 20 km) and then downscaled statistically to 8 km resolution. Our results showed that simulations at fine resolution had relatively little impact on the mean and variance of beech productivity over France. However, at local scales, the mean and variance of beech productivity changed strongly (up to 60%) as a result of finer scale simulations, particularly in regions characterized by complex orography such as the Alps or the Mediterranean. Our results highlight the need to consider the importance of spatial scale of climate input when studying climate impacts on tree distributions in areas with steep climatic gradients.