Dear Prof. Savenije,

As the main goal of opinion papers is to stimulate a vital and thoughtful debate on key topics I am, given the thoughtful contributions to the discussion, very happy to host this one HESS. W.R. Berghuijs, M. Cuthbert and the reviewer Stefan Hergarten did an excellent job in reviewing and criticizing the proposed derivation of the linear reservoir from Darcy’s law. And I very much agree with all of them that your responses their comments very much clarified the underlying thoughts of the proposed theory. I thus encourage you to moderately revise your work as outlined in your responses. During this effort you might also consider find the following points:

I think the paper could win from being a little bit more precise about a) where and when we observe exponential recessions and b) where we use the linear reservoir in modelling. Allow me to explain:
It is correct that recessions are often not entirely linear. But what is also true is that the ratio dQ/dt over Q gets often constant at later stages of the recession, which implies that base flow recession gets more and more in accordance with the linear reservoir (Blume et al. 2007, sorry for citing my own work). And it is thus only base flow recession we usually mimic with a linear reservoir in conceptual models. These models contain in addition non linear storage elements to simulate a) earlier non-linear stages of the recession and b) runoff formation itself.

The interesting question is hence maybe not whether and why catchments behave like linear reservoirs, but more precisely why catchments do start draining similar to a linear reservoir at some stage. In this context it is important to recall that we observe this exponential behavior not in the ground water store itself (e.g. from head data) but we observe it in stream flow at the catchment outlet. So in fact the exponential recession is the ‘end product’ of the exchange between the groundwater/aquifer system and the stream in form of base flow generation and downstream flow in the channel to the catchment outlet.

Can we thus understand this process without taking care for the channel? This leave us with the question what is limiting this behavior? Is it the aquifer and it properties or is it the channel and the development of its geometry downstream? I am not sure....

If we assume for instance a constant base flow production per unit channel length (maybe too simple), this implies a steady linear increase of river discharge along the channel. When assuming a constant flow velocity, the river has for instance to widen downstream in a sufficient manner to accommodate the added water and yet to maintain the driving water level gradient to assure steady state flow along the channel. So is it the aquifer with it’s embedded preferential pathways or is the channel (as preferential pathway) and the downstream development of the width function which is explains why exponential recessions emerge?

Last not least I like to stress that your assumption of the constant resistance does not at all neglect variability in permeability. On the contrary it implies that permeability increases with increasing flow path length to the channel. This in turn implies a rather strong organization of permeability in the catchment.

Best regards,

Erwin Zehe

References:
Blume, T., Zehe, E., and Bronstert, A.: Rainfall-runoff response, event-based runoff coefficients and hydrograph separation, Hydrological Sciences Journal-Journal Des Sciences Hydrologiques, 52, 843-862, 10.1623/hysj.52.5.843, 2007.

Dear Dr. Savenije,

I am delighted to accept your opinion papers for final publication in HESS. As you can see, both reviewers are very pleased by the way how you adressed their comments ... and so am I.

Best regards,

Erwin Zehe