As indicated by Dr Teuling (and his students) and the anonymous reviewer a major revision is required. Apart from the points that you already addressed in your replies to the reviewers, particularly the following points require some more elaboration before the paper is suitable for publication in HESS:
• make sure you reference to existing literature that has the same application/definition of root soil moisture as you generate in your study
• apply metrics to anomalies (“deseasonalized”, as you call it in your reply to reviewer 1), and include a measure of the statistical significance of these metrics
• justify the use of a model data set such as FLEX as benchmark for your evaluation
• discuss (and possibly correct) the assymetry in correspondence between Su and NDII for wet and dry conditions
• consider to include a clearly formulated hypothesis (a question that can be answered by “yes” or “no”

Dear authors

A new reviewer has been invoked to evaluate this resubmitted version of your manuscript, and this reviewer recommends again a major revision. I do agree with this qualification, since also my recommendation to build up the paper along a clearly testable research hypothesis is not implemented very well in this revised version. Also the reviewer points at lack of adequate response to points raised earlier, particularly concerning the clarity of your explanation, and the solidness of the conclusions.

I therefore give you another opportunity to revise the paper accordingly, and will send this new revision out to this reviewer. I will also allow Dr Ryan Teuling to review this next revision.

The new review and my editorial comments have pushed the paper into a direction that is indeed better than the first submission, but does not address the concerns in a way that is convincing enough for accepting the manuscript in its present form. To my opinion there is no doubt about the scientific value of this comparison between NDII and FLEX output, but the framing that NDII can be considered a reliable proxy of soil moisture even in ungauged basins is not adequately demonstrated by this study. The formulation of the hypothesis does not include any verifiable quantitative assessment (like “the NDII method is better than…” or “the NDII method is good enough to…”), and the current study set-up does not allow to test such a quantitative hypothesis.

However, I do agree with the authors that the study in itself is interesting enough to be published. So rather than pushing this study towards this prove of adequacy of NDII as a soil moisture proxy in ungauged basins, I would recommend to downplay the framing and concentrate on the basic results: comparing NDII to a model derived soil moisture index does lead to reasonable correspondence and gives scientific insights that support the value of NDII (such as the effect that it only works well under dry conditions, and does discriminate between different subcatchment typologies). One step to bring the expectations closer to the true value of this study is to change the title into something like “Comparison between NDII and results from a lumped hydrological model”.

Having said that, the technical layout of the manuscript is still not good enough. Not only is the supplementary information (showing results for the other basins than P.20 and P.21) is missing and is the format of the manuscript file(s) very inconvenient, it also contains quite some editorial issues (unclear cross-referencing to equations, an unclear description of the spatial set-up of the FLEX model in section 3.2, arguments for using the current study set-up that are distributed across various sections rather than concentrated in the introduction section).

The new reviewer has requested to evaluate a new version of the manuscript. In that review a much more thorough quantitative evaluation (with a distributed model, in ungauged basins) was required for this manuscript to be publishable. I do not insist on these new model simulations and analyses if the authors are able to improve the framing of this study and concentrate on the (valuable) findings that were found in their analyses. So I will enable the autors to submit a new version of the manuscript that addresses the points raised above without sending it out for review at this stage. However, this new manuscript will be sent out for review if that appears appropriate at that time.

Dear authors, the somewhat a-typical review process (involving students, new reviewers at a later stage) has led to a paper that clearly shows the potential benefit of using remotely sensed information on vegetation status to diagnose the hydrological state of a river (sub-)catchment. Although more advanced analyses with the data could have highlighted the method's ability to discreminate between seasons, basins and other conditions a bit more, the evaluation presented in this paper is technically sound, and supports the conclusions drawn. Therefore I am happy to let you know that the paper can be published in its present form.