Transpiration dynamics of an Austrian Pine stand and its forest floor: identifying controlling conditions using artificial neural networks

In this study, artificial neural network analyses (ANN) were used to identify the forcing environmental variables that are most significant in governing the transpiration rates of an Austrian Pine stand and its forest floor. Latent heat flux densities (Lh) of the Austrian Pine stand and its forest floor were separately measured using the eddy covariance technique. To assess the sensitivity of the ANNs to input information on the soil water status, the site calibrated soil hydrological model SWIF was used to compute average volumetric soil water contents of different depth intervals. Results show that forest floor transpiration dynamics can be adequately modelled using the global radiation reaching the forest floor and the topsoil water content (0–50 cm). The response functions of the total forest and forest floor showed a clear difference in sensitivity of latent heat fluxes to global radiation, air temperature and soil water content. Most significantly, results demonstrate that the presented ANN analysis is suited for assessing effective rooting depths from measured transpiration rates and soil water contents.