Hydrological models account for the storage, flow of water and water balance in a catchment, including exchanges of water and energy with the ground, atmosphere and oceans. Because of the need to simplify hydrological models, parameters are often included to help with modelling hydrological processes. Generally the parameters of lumped, semi or distributed hydrological models depend on the values estimated at a gauged location, generally at outlet of the catchment (mostly using discharge). In this approach it is hard to judge how well a model represents internal catchment processes of the hydrological cycle as well as also ignoring the spatial heterogeneity of the catchment. The purpose of the paper is to assess, and potentially improve, the ability of a physically-based semi-distributed hydrological model, TopNet, using a spatially and temporally detailed set of field measurements of catchment responses to diverse weather conditions. The TopNet rainfall-runoff model was applied to the Waipara catchment located in the South Island of New Zealand. Observations from field experiments were compared with the simulation results of uncalibrated TopNet model. The total amount of simulated runoff showed reasonable agreement with observations. However, the model overestimated baseflow and underestimated surface flow. The results show that soil moisture variation within the catchment is well represented by the model. However, comparison of the observed water balance with model results show that there is a deficiency in the calculation of evapotranspiration.