The 11 papers appearing in this special issue of Hydrological Processes focus on the 'downward' or 'topdown' approach to hydrological prediction. The downward approach was introduced into the hydrologic literature by Klemeŝ (1983) who defined it as the 'route that starts with trying to find a distinct conceptual node directly at the level of interest (or higher) and then looks for the steps that could have led it from a lower level'. The downward approach is an empirical or data-based approach, involving interpreting or learning about a catchment's functioning from data obtained at the catchment scale, or from patterns in the observed data. Jarvis (1993) clarifies that any such explanation and/or generalization (of what is seen in the data) is achieved by fingering down into the (smaller scale) processes from above (i.e. catchment scale).
In spite of its obvious potential, the downward approach is a relatively new phenomenon in hydrology. We strongly believe, however, that it will, in time, become an important component of the toolkit for data analysis and hydrological model development. In particular, it will serve as a necessary counterpoint to the process-based 'reductionist' approach that dominates current hydrological model development. Indeed, we believe that progress in hydrological prediction will come from some form of synthesis of the traditional upward or reductionist approach (based on basic laws of physics) with the downward approach (based on systematic learning from data).
This special issue is an outgrowth of a special workshop held in Melbourne, Australia, in October 2000, under the sponsorship of the Cooperative Research Centre for Catchment Hydrology, for which we are extremely grateful. We thank the many scientists who participated in the discussions at the workshop and the authors who contributed articles to this special journal issue. In particular, we thank the following reviewers, from around the world, who assisted us in providing critical reviews of the manuscripts: