Models underpin all numerical analysis that is undertaken in catchment studies. They can take many different forms, from simple empirical relationships to complex three-dimensional spatially distributed representations of transport processes. In all cases, the models are an imperfect representation of a system constructed on the basis of experimental data and an imperfect understanding of processes. The experimental data themselves provide an imperfect representation of the real world, limited by sample frequency, spatial resolution, and measurement accuracy. Thus, the success of model applications depends not only on the model structure but also the appropriateness of the data on which the model is based, and, importantly, the purpose for which the model is required. This Commentary resulted from a workshop in Ballater, Scotland in May 2007, on 'From Catchment Scale Process Conceptualisation to Predictive Capability' (summary at http://www.abdn.ac.uk/โผwpg027/w shop.php), which sought to integrate new insights from process conceptualization (Tetzlaff et al., 2008) and catchment data (Soulsby et al., 2008) into modelling studies, and show how, in turn, modelling can help focus field-based investigations. Extensive recent hydrological modelling literature has been devoted rightly to considerations of uncertainty. This has advanced from early debates relating to parameter uncertainty (Beven and Binley, 1992), to a much broader consideration of end-to-end uncertainty (Pappenberger et al., 2005) incorporating input data (Tetzlaff and Uhlenbrook, 2005; Kavetski et al., 2006), model structure (Son and Sivapalan, 2007) and validation data (Harmel and Smith, 2007
). Although such discussions are essential and valuable, the consequence has been the perceived establishment of a barrier to the use of models in applied catchment hydrology, as the uncertainties frequently cast doubt on the interpretation and validity of model results. These issues need to be embraced for the future, through the promotion of an appropriate modelling culture that recognizes the limitations of models but that can take advantage of their intrinsic value in formalizing our understanding of the complex inter-relationships among processes, parameters, and places.
Models, used to complement our interpretation of catchment processes, may need to be integrated in catchment studies in an iterative format that involves several cycles of development and evaluation. Where model analysis can be carried out in parallel with other approaches, such as field-based data collection, there is the potential to provide multidirectional synergistic feedback between the approaches. Involvement of the end-users of a catchment study throughout the procedure may lead to the provision of additional 'local' knowledge, as well as ensuring that the focus of the study will be appropriate to their needs. A procedure of this type can be considered as a form of learning framework, such as that illustrated in Figure 1. Here we will consider how conceptual catchment modelling can play a key role in this type of approach, and the required procedures and infrastructure that will be necessary for its establishment.