Modelling evapotranspiration in a scots pine stand under Mediterranean mountain climate using the GLUE methodology.

摘要

Canopy transpiration (E< sub>c) and soil evaporation (E< sub>s) in a Mediterranean Scots pine stand were simulated using a two-layer model, with a Jarvis-type submodel of canopy stomatal conductance (G< sub>s) and a soil resistance to evaporation expressed as a function of superficial soil moisture. Sap flow measurements and soil evaporation data, together with meteorological and soil moisture variables were used to calibrate the model. G< sub>s was calibrated using the generalized likelihood uncertainty estimation (GLUE) methodology, first with data from the year 2004, a year characterised by mild meteorological conditions. Then, data from the year 2003, which included an intense summer drought, was used to update the results from the previous calibration. The discrepancy between the diurnal courses of modelled and measured E< sub>c using best-fit parameters was not related to any particular situation of meteorology or soil moisture. Model performance improved at the daily scale, but the model failed to simulate E< sub>c adequately during the year 2005. Maximum modelled E< sub>s rates were 0.7 mm day-1 with the ratio E< sub>s/E< sub>c being typically under 0.3 during the growing season. The GLUE analysis revealed that parameters representing reference stomatal aperture at a vapour pressure deficit (D) value equal to 1 kPa (G< sub>s,ref), and sensitivity to D (m) were the most relevant, and were consistent with the hydraulic theory of stomatal regulation. Parameters controlling the response to superficial soil moisture deficit only appeared sensitive in the calibration with data from the year 2003, suggesting that response to deeper soil layers should also be considered in the model. Updating the original calibration reduced predictive uncertainty and constrained the value of some parameters. Nevertheless, it seems that representations of variable plant and soil hydraulic resistances, are required to simulate long-term E< sub>c in seasonally-dry Mediterranean forest stands.