Modelling surface conductance and transpiration of an oak forest in The Netherlands

Abstract

The surface conductance of an oak forest was modelled as a non-linear function of solar radiation, temperature, specific humidity deficit, soil moisture deficit and leaf area index. Hourly values of the surface conductance were derived from the Penman-Monteith equation, using transpiration measurements of a red oak (Quercus rubra) forest in the Netherlands. The transpiration was determined from the energy balance and Bowen ratio measurements. A total number of 576 hourly measurements were available, of which 162 were in 1988 and 414 in 1989.The piecewise linear specific humidity function in the Jarvis-Stewart model (model A) was adapted in this study to an exponential form. For this modified model (model B), average values of simulated hourly surface conductance, using independent data sets for parameter estimation and simulation, differed by less than 3% from measured values. The difference between simulated and measured average hourly transpiration was less than 5%. The model seemed to overestimate the surface conductance in the late afternoon, but the difference between hourly averaged simulated and measured surface conductance was not significant. The sensitivity of model B was also investigated. The model appeared to be most sensitive to variations in the maximum surface conductance and in the parameters of the specific humidity deficit function. Replacing the exponential function of specific humidity deficit by a piecewise linear function did change the estimated parameters considerably. Simulation results for the model with the exponential specific humidity function (model B) were better than for the model with the linear function (model A). When parameter estimations for model A, obtained at pine forests in the UK and France, were used for simulation of surface conductance of the oak forest, the difference between average simulated and measured values was −22% and 13%, respectively. Transpiration differed by −23% and 15%, respectively.In conclusion, both model A and model B gave a good estimation of average surface conductance and transpiration, but care should be taken when using parameter values obtained at one location for predictions at another location. For comparison of parameter values obtained from different data sets, it is essential that exactly the same algebraic formulation is used.