The aim of this work was to test a process-based model (hydrological model combined with forest growth model) on the simulation of seasonal variability of evapotranspiration (FT) in an even-aged boreal Scots pine (Pinus sylvestris L) stand over a 10 year period (1999-2008). The water flux components (including canopy transpiration (E(t)) and evaporation from canopy (E(c)) and ground surface (E(g)) were estimated in order to output the long-term stand water budget considering the interaction between climate variations and stand development. For validation, half-hourly data on eddy water vapor fluxes were measured during the 10 growing seasons (May-September). The model predicted well the seasonal course of ET compared to the measured values, but slightly underestimated the water fluxes both in non-drought and drought (2000, 2003 and 2006) years. The prediction accuracy was, on average, higher in drought years. The simulated ET over the 10 years explained, on average, 58% of the daily variations and 84% of the monthly amount of E. Water amount from E(t) contributed most to the ET, with the fractions of E(t), E(c) and E(g) being, on average, 67, 11 and 23% over the 10-year period, respectively. Regardless of weather conditions, the daily ET was strongly dependent on air temperature (T(a)) and vapor pressure deficit (D(a)), but less dependent on soil moisture (W(s)). On cloudy and rainy days, there was a non-linear relationship between the ET and solar radiation (R(o)). During drought years, the model predicted lower daily canopy stomatal conductance (g(cs)) compared with non-drought years, leading to a lower level of E(t). The modeled daily g(cs) responded well to Da and W. In the model simulation, the annual LAI increased by 35% between 1999 and 2008. The ratio of E(c):ET correlated strongly with LAI. Furthermore, LAI reduced the proportion of E(g) as a result of the increased share of E(c) and E(t) and radiation interception. Although the increase of LAI affected positively E(t), the contribution of E(t) in ET was not significantly correlated with LAI. To conclude, although the model predicted reasonably well the seasonal course of ET, the calculation time steps of different processes in the model should be homogenized in the future to increase the prediction accuracy. (C) 2011 Elsevier B.V. All rights reserved.
