Abstract
The main objective of this study is to identify runoff generation mechanisms in a forested catchment (0Β·18 km^2^) by comparing model simulations against measurements of groundwater levels and streamflow, and against separation of runoff components based on isotope tracer analysis. The hydrological model used in this study had a physical basis, and it consisted of separate procedures for canopy, snow, and hillslope processes, and for routing in the channel network. This paper concentrates on calibration and validation of the hillslope and routing procedures. Winter and summer seasons were examined separately in order to isolate different mechanisms controlling groundwater level behaviour. The results indicated that seasonal runoff dynamics could be replicated without calibration of the hillslope procedure against measured streamflow. The routing procedure could merely delay the runoff input it received from the hillslope procedure, and it had no influence on how large a share of a throughfall/snowmelt event, under different preceding climatic conditions, eventually emerged as streamflow. Comparison against isotope tracer data suggested that use of groundwater level measurements in the model calibration could lead to a more plausible simulation of new and old water contributions to streamflow. Still, the modelled new water fraction in streamflow was nearly twice as large as the fraction derived from isotope tracer analysis. Improvement of the model structure towards a more realistic separation of runoff components would require the model to consider residence times of water in different parts of the soil domain. Copyright Β© 2003 John Wiley & Sons, Ltd.