Dear authors,

thank you very much for your thoughtful replies to the reviewer comments.
I think that you addressed most important points raised in a detailed and adequate way. I do only have a few minor remarks, which I would encourage the authors to consider in addition to the reviewer comments:

(1) while it is true that tritium does carry quite some information content about the age of water, I think the often communicated "one sample is enough" message needs to be toned down a bit (p.5,l.6). of course one sample is better than no sample, but given the observational and modelling uncertainties involved, it will only give us a very rough estimate. in principle, models could be calibrated using one O-18 value as well. the question is then, however, how much confidence do we have in this one sample and how strongly can it constrain redundant parameterizations? the same is of course true for tritium.

(2) in your replies you mention that MTT studies in peatlands are rare. this is not exactly true. most of the british MTT work (e.g. soulsby, tetzlaff, birkel and myself) is located in peatlands.

(3) in one comment reviewer 2 highlighted the missing description of the model calibration strategy - i.e. how were the parameters determined? what were the objective functions used? what was the calibrated model performance? you did not address this point in your replies: you only mention your assumptions on the choice of the model but not on how you obtained the parameters. please add that in the revised manuscript.

(4) p.6,l.11-17 would indeed better fit into the methods section as suggested by reviewer 2

(5) apart from that i think you can leave the introduction section and its sub-sections structurally as is.

i am very much looking forward to receiving a revised version of your manuscript!
best regards, markus hrachowitz

Dear authors,

thank you very much for your detailed responses and adaptations. The paper reads really good now, but I would be glad if you could address two remaining minor points, before I accept the manuscript for publication:

(1) I strongly disagree with your statement that what you refer to as "time series approach" to estimate transit time distributions and mean transit times requires the assumption of time invariance. It is true that most lumped, convolution integral models work that way. Yet, this is not true for all; see for example Weiler et al., 2003, who introduce weighing factors to account for variable flows, or of course the original work by Niemi (1977) and his concept of flow time. In addition, there was considerable progress over the past years, explicitly allowing for variable flow conditions also in lumped models (e.g. Botter et al., 2010,2011; Benettin et al., 2013; Rinaldo et al., 2015).

(2) I agree that testing several models is a good idea when insufficient information about the actual system is available. This, however, does not answer the question raised by Reviewer #2: How did you choose for the different functional shapes of transit time distributions (e.g. epm, em, dm) the associated transit times (tau or taum) which you need in your eqs.1-3. This is not clear from the text and needs to be explained to allow the reader to completely follow what you are doing, because surely different values of tau can produce the observed signal with different recharge activity and different values for the other parameters. In other words, as far as I understand, you vary the recharge H3 activity and the other parameters but not tau. This needs to be clarified and justified.

best regards,
Markus Hrachowitz

Dear authors,

thank you very much for the final corrections! I enjoyed reading the revised manuscript, which is very interesting and which can be published as is.
Congratulations!
Best regards, Markus Hrachowitz