Exploring the interplay between state, structure and runoff behavior of lower mesoscale catchments

Abstract

The question of how catchments actually "function" has probably caused many sleepless nights as it is still an unsolved and challenging scientific question. Here, we approach this question from the similarity perspective. Instead of comparing single physiographic features of individual catchments we explore the interplay of state and structure on different runoff formation processes, aiming to infer information on the underlying "functional" behaviour. Therefore, we treat catchments as lumped terrestrial filters and relate a set of different structure and storage descriptors to selected response measures. The key issue here is that we employ dimensionless quantities exclusively by normalizing the variable of interest by its limiting terrestrial or forcing characteristic. Specifically we distinguish extensive/additive and intensive/non-additive attributes through normalizing storage volumes by maximum storage capacities and normalizing fluxes (e.g. discharge) by permeability estimators. Moreover, we propose the normalized temporal derivative of runoff as a suitable measure to detect intensity-triggered (high frequency) runoff production. Our dimensionless signatures evidently detect functional similarity among different sites for baseflow production, storm runoff production and the seasonal water balance. Particularly in the latter case we show that normalized double and triple mass curves expose a typical shape with a regime shift that is clearly controlled by the onset and the end of the vegetation period which we can adequately characterize by a simple temperature index model. In line with this, temperature explained 70 % of the variability of the seasonal summer runoff coefficients in 22 catchments distributed along a strong physiographic and climatic gradient in the German part of the Danube basin. The proposed non-additive response measure detected signals of high frequency intensity controlled runoff generation processes in two alpine settings. The approach, in fact edge filtering, evidently works when using "low-pass" filtered hourly rainfall-runoff data of mesoscale catchments ranging from 12 to 170 km2. We conclude that vegetation exerts a first order control on summer stream flow generation when the onset and termination of summer are more significantly defined by temperature than simply by the actual Gregorian day. We also provide evidence that properties describing gradients (e.g. surface topography) and resistances (e.g. hydraulic conductivities) may be much more powerful in explaining runoff response behaviour when they are treated as groups compared to their individual use. Lastly, we show that storage estimators such as the proposed normalized versions of pre-event discharge and antecedent moisture can be valuable predictors for event runoff coefficients: For some of our test regions they explain up to 70 % of their variability.