MULTI-YEAR SIMULATION OF FOREST CARBON FLUXES IN ITALY BY THE COMBINATION OF GROUND AND REMOTELY SENSED DATA

Quantifying the amount of carbon stored within forest ecosystems is a really important task in view of the global climatic changes and under the broader perspective of ecosystem service assessment. Among the methodologies proposed to obtain such quantification (e.g. eddy covariance techniques, remote sensing and bio-geochemical models), those based on the integrated use of models of ecosystem processes and remotely sensed data are the most promising. An integration methodology based on two models, modified C-Fix and BIOME-BGC, has been currently applied to predict both gross and net carbon fluxes of forest ecosystems in Italy. In particular, C-Fix is a Monteith model which combines meteorological and Normalized Difference Vegetation Index (NDVI) data to predict forest production (GPP), and BIOME-BGC is a bio-geochemical model capable of estimating photosynthesis, respirations and allocations. The outputs of these two models are integrated by the use of a proxy of the distance from quasi-equilibrium condition, which is the ratio between actual and maximum simulated standing volume. These two models were applied after the collection of all input data at a proper spatial and temporal scale. In particular, the daily meteorological data (both minimum and maximum temperature and precipitation) were obtained by a pan-European dataset (E-OBS) which provides them at 25 km spatial resolution. These data were then downscaled to 1-km spatial resolution applying locally calibrated regressions to a digital elevation model. Solar radiation was then estimated by the algorithm MT-Clim. Forest distribution was derived from the CORINE Land Cover 2006 map, while information on the growing stock was derived from a wall-to-wall map produced using the global canopy height map derived from Geoscience Laser Altimeter System (GLAS). NDVI images were taken from the Spot-VEGETATION satellite sensor and utilised to produce the fAPAR estimates required by C-Fix. All these data are available for a fifteen-year period (1999-2013). The general validity of the whole modelling approach was demonstrated against independent data collected from different sources: measurements collected during regional and national forest inventories, and/or data collected by the eddy correlation technique (both GPP and NEE). The results of these comparisons were generally positive, indicating the good capability of the method for the estimation of forest carbon fluxes in Mediterranean areas. More specifically the NDVI driven C-Fix model is capable to produce better GPP estimates than BIOME-BGC; concerning net forest fluxes, the best results are obtained when the C-Fix GPP estimates are integrated with the estimates of BIOME-BGC respirations. In the latter case, the correction for quasi-equilibrium condition is fundamental to obtain realistic estimates of actual forest fluxes. After these validation activities, maps of the main forest carbon fluxes (i.e. GPP, NPP and NEE) were produced for the whole study period and the effect of dry/wet years on forest ecosystems was investigated.