Scaling of dominant runoff generation processes: Nested catchments approach using multiple tracers

Low-flow stream chemistry and rainfall event reactions of nested catchments in the mesoscale Dreisam catchment in the Black Forest Mountains, southwest Germany, were analyzed to investigate how the dominant runoff generation processes change with scale. The catchment sizes range from a 0.015 km(2) headwater catchment to the 258 km(2) Dreisam catchment. Synoptic sampling during low flows was used to address the spatial heterogeneity of the investigated tracers. Six events were sampled using three different experimental designs with the environmental tracers dissolved silica, oxygen-18, deuterium, and potassium to investigate event processes at different scales. Results are elucidated with respect to the widely noticed Representative Elementary Area (REA) concept. Most of the observed differences between the catchments could be related to changes in the topography and other catchment properties (i.e., soils, geology, land use). The test site catchments less than 1-2 km(2) were found to be nonrepresentative of the runoff generation in larger catchments due to the topographic structure and a reduced number of hydrological response units (HRU) and, consequently, generated runoff components. In catchments larger than 40 km(2) an additional runoff component, the surface runoff from urban areas, became increasingly important. However, surprisingly small differences in the tracer responses at catchments between 1 and 40 km(2) were observed. Although the lower threshold (1-2 km(2)) was similar for both methods (low-flow and event investigations), results suggest that the thresholds depend on the investigated scale and hydrological parameters as well as the catchment properties. Applying microscale tracer methods at the mesoscale provided detailed insights into the scaling of the dominant runoff processes. The results show that this approach is an important component when addressing the scaling behavior besides the numerous microscale studies and modeling approaches. However, quantitative interpretations are limited owing to inherent heterogeneity at this scale.