Application of the 3-PGS model to assess carbon accumulation in forest ecosystems at a regional level.

In this study we assessed carbon sequestration by Italian forest ecosystems at a regional level. We applied a monthly time-step process-based model (3-PGS), coupled with a modified soil respiration model, to predict both gross primary production (GPP(3)-PGS) and net ecosystem production (NEP(3-PGS)). To evaluate the general reliability of model estimates, we compared, at five different forest sites, monthly and annual GPP(3-PGS), NEP(3-PGS), and predicted total ecosystem respiration (TER(3-PGS)) with averages of monthly and annual eddy covariance (EC) measures of GPP(EC), NEP(EC), and TER(EC). A strong correlation was found between annual GPP(3-PGS) and annual GPP(EC) (r(2) = 0.77, RMSE = 1.28 Mg C.ha(-1).year(-1)), and monthly (r(2) = 0.85, RMSE = 35 g C.m(-2).month(-1)), as well as between NEP(3-PGS) and annual NEP(EC) (r(2)= 0.76, RMSE = 0.21 Mg C.ha(-1).year(-1)), and monthly (r(2) = 0.78, RMSE = 18 g C.m(-2).month(-1)). The TER(3-PGS) also showed a high correlation with annual TER(EC) (r(2) = 0.93). Furthermore, a sensitivity analysis showed that GPP(3-PGS) was highly sensitive to the satellite greenness index (normalized difference of vegetation index) and to the vapor pressure deficit. With general confidence in the models, we established a 30 year average meteorological grid of 8 km * 8 km resolution across Italy and created a map representing annual NEP(3-PGS) across Italian forests, based on the remotely sensed CORINE Land Cover forest classification.