and UNESCO and was attended by nearly 100 participants from 16 different countries. A number of sessions over 4 days explored the themes of: Model design, scale and parameter estimation issues; Modelling the impacts of land use and climate change; Remote sensing and GIS in distributed modelling; and Soil erosion modelling [see also the commentary on the meeting in HPToday in Hydrological Processes 2000; 14(16-17): 3183].
Each theme had two invited speakers and further oral and poster presentations to set the scene for the most important part of the meeting: the discussions, which were appropriately aided by some excellent Belgian beers (the Stella brewery in Leuven dates back even longer than the university, even if it is now owned by Interbrew).
The standard of the presentations and discussions at the meeting was very high, and this is reflected in the papers prepared after the meeting and accepted for this issue. Thanks are due to the very many referees whose careful and critical reviews were very important in significantly improving the final versions of some papers. The papers span the range of the very important issues that must be addressed in the future of distributed hydrological modelling: of how best to represent hydrological processes at different scales; of how best to identify the parameters of distributed models at different scales lacking information and measurement techniques; and of how best to incorporate new sources of data, especially from remote sensing, into the modelling process at different scales? There are also some applications of different model formulations to a wide range of catchments in different environments and some explorations of the role of spatial variability of parameters on model predictions.
If the reader perceives that these papers reveal the problems to be addressed in the application of distributed models more than providing solutions, then that properly reflects the current state of the art. At least, as a community, we seem to appreciate the nature of the problems involved far better than 10 years ago, when the first workshop on the future of distributed modelling was held at Lancaster University as a British Hydrological Society national meeting [see papers in Hydrological Processes 1992; 6(3).] Appreciation of the problems is, at least, a start towards a solution. So what will be the state of the art if there is another workshop in 10 years time?
Looking back to the preface of the special issue in 1992, no predictions were made about what would happen in the next 10 years. In fact, not a great deal actually seems to have changed since 1992. Certainly distributed models have become easier to use and visualization of the results has greatly improved, but there has been very little change in the concepts on which the models are based and the ways in which they are calibrated and used. Should we hope for more change in the next decade? We can certainly expect that the computer power available to the distributed modeller will continue to increase. Cheap parallel computers are already becoming widespread, and the operating systems and compilers available to make simple use of them should become greatly improved. Visualization techniques for the presentation of 4D simulation