Future of distributed modelling

The state of the art of modelling hydrological systems using distributed modelling techniques is discussed and the theoretical problems caused by the heterogeneities always present in field soil systems are emphasized. It is argued that such heterogeneities need to be considered to develop physically based large‐scale models applicable at the scale of the numerical discretization blocks. Stochastic approaches seem to provide a rational framework for treating the difficult to characterize spatial variability and for establishing a link between parameter uncertainty and prediction uncertainty. It is emphasized that more research in theory development and testing is required before complete large‐scale models are obtained and this had to be carried out jointly with field tests. More research is also required in field measurement techniques to accurately supply the deterministic and statistical parameters required in distributed models.