Identifying hydrological regularities via perceptual models at the regional scale

Identifying hydrological regularities, such as patterns and laws that explain the observed variability in catchment response, is an important objective of catchment hydrology. These insights could contribute regional knowledge that can be exploited in catchment classification studies and improve model realism and parsimony. But how to infer such regularities, and to what extent are they generalizable, in view of the evidence of uniqueness of place? In this presentation, we propose the development of a regional scale perceptual model as a framework to stimulate the search of hydrological regularities and to represent them visually. Our approach is intended for nested catchments, a scale that we consider is sufficiently large to provide an interesting contrast in hydrological responses, but sufficiently small to encompass local dominant process that may be different elsewhere. Our perceptual model development approach is demonstrated in the 27,000 km² Moselle catchment, using streamflow data at 26 nested subcatchments, and commonly available data of landscape properties, including topography, vegetation, geology and soil. The identified signatures of streamflow spatial variability highlighted the role of precipitation, geology and topography, which affected, respectively the average flows, base runoff and lag time. Soil and vegetation, on the other hand, were not found to be a dominant cause of hydrograph variability, which might appear surprising, considering that soil properties are one of the key ingredients of many distributed models. The framework undertaken in this study may be useful to develop perceptual models in other basins at regional scale, and to map and regularize the variety of dominant hydrological processes.