Hydrologic modeling and forecasting: role of thresholds

This communication discusses the potential role of 'thresholds' in hydrologic modeling and forecasting. Based on representative examples of studies employing data learning techniques in hydrology (e.g. artificial neural networks, chaos theory, optimization procedures), some of the problems in the existing data calibration and validation procedures are highlighted. Deriving an analogy between human behavior and catchment behavior, emphasizing their internal characteristics and their responses to external events, the concept of thresholds and its use in catchment hydrology are presented. In view of recent calls to focus on dominant processes in hydrology, a proposal is made for integration of concepts of Dominant Processes and Thresholds. The promises and limitations of this proposal are also explained in brief.