|Review of “A promising new baseflow method and recession approach for streamflow at Glandhu Catchment, New Zealand” by M.K. Stewart|
This is a much revised manuscript following earlier reviewer and editor comments. I found the revised manuscript somewhat improved in structure, but several oddities remain (e.g. subsection 2.2.1 begins mid-paragraph “However,……”). Despite the restructuring and new sections that have made this a somewhat different paper with a modified title, the author has not addressed the fundamental problem with the manuscript: the tone of presentation in the paper. It remains a set of sweeping statements that dismiss recession analysis techniques and baseflow separation as being “in disarray”, and then draws broad conclusions based on two hydrograph events – a total of only 16 days of data from a 2.18 km2 watershed, and declares very simple empirical model as being “more accurate” than previous methods, but compares this model with a very limited subset of the available approaches, extols the few advantages of the new method but grossly underplays its limitations. It is not reasonable to evaluate the general utility of an empirical model that was derived from two hydrograph events, in a very small watershed, by measuring that model’s ability to reproduce the very responses from which it was derived!
The manuscript’s sections that focus on the field site and analysis of the data are much improved in terms of clarity, but the presentation of the results, is severely compromised by the framing of the manuscripts central arguments as set out in the abstract and introduction. Broadly speaking:
(1) That there is no general consensus on how to approach the problem of baseflow and recession analysis, even though there are many different methods available for each;
The author does not really provide a thorough review of this area, but makes sweeping statements based on selected quotes from previous papers, many of which were written 20 to 50 years ago when methods were considerably less advanced than they are now. Because of this, many key references are missing, and there is no detailed consideration of the methods – yes, the methodology states some of the approaches, but the treatment is not in any depth. The author criticizes a lack of consistency in approaches, but does not provide the detailed level of analysis to illustrate what distinguishes the different methods – what assumptions underpin each approach, why are they necessary, and what is their potential impact on the results? All analysis of baseflow and recession curves requires a set of assumptions, and the results will depend on what assumptions were made. The introduction doesn’t provide anything but a very selective overview that is highly dismissive of the available approaches and fails to consider the technical background that would allow a meaningful discussion of differences to be made. These opening paragraphs have an effect that totally undermines the potentially interesting results from Glendhu, because it frames the whole manuscript as the search for a solution to the question of how to do baseflow and recession analysis, rather than using them as a way to demonstrate how a particular model might be used in particular circumstances.
The present manuscript now states several methods, but it does not really evaluate them in a critical and meaningful way that would allow a balanced evaluations of their relative merits and the uncertainties likely to result in any analyses or subsequent interpretation. Without such a detailed view it is simply impossible to say that one approach is “better” than any another.
(2) That performing recession analysis prior to baseflow separation can be highly misleading;
Can it – I am sure it can, but under what conditions, in what sort of watersheds, for what kind of events? That sounds like a very interesting question! This is simply stated by the author – and I expected it to be followed by a detailed analysis to prove the point, but there is no evidence or illustrative discussion presented to support the assertion. This is simply the author’s opinion and is stated on line 91 “whilst this may be considered obvious by some”… This statement appears to fundamentally misunderstand why a mathematical analysis is done in this order – but the logic is carefully described by Eckhardt (2005), such that the comments in lines 91-92 just appear naïve and dismissive. If the author wants to advance the notion that such procedure is a big problem, then he must provide evidence for why that is the case – not simply state that it is obvious, because it frankly it isn’t. I agree that a constraint of digital filters is that several assumptions must be made, one of which enables the separation of baseflow from rapid runoff through the identification of a baseflow recession parameter during a period of assumed zero rapid runoff, by selecting a period when there is no incident precipitation, but when the streamflow continues to recess. It is perfectly valid to question that assumption, but to do that requires empirical evidence or a theoretical model to explain why this approach is subject to uncertainty!
(3) that applying baseflow separation analysis before recession analysis can resolve some problems with recession analysis and provide new and important insights into catchment water storages; and,
This point follows from (2) above. But again, this is just stated, without any evaluation of a substantial body of evidence to support the statement – except 16 days of data from Glendhu! I don’t disagree that there are potential problems with performing the analysis in the way that most digital filters do, but to argue this point convincingly requires a very detailed analysis of the way in which the underlying assumptions impact on the resulting analysis- which is not presented.
(4) the new “bump and rise” baseflow separation method provides “more accurate” baseflow simulations than previous methods.
The manuscript uses two sets of sampling data – one from a 2-day period in 1988, and one from a 14-day period in 1996 – and uses these, along with the simplest of all baseflow separation methods developed in the late 1960s (the Hewlett and Hibbert (1967) approach) to develop the “bump and rise” method for the Glendhu catchment – area 2.18 km2. The results of this new model, derived from 16 days of observations in a 2.18 km2 watershed are then declared as universal truths – the conclusions do not frame the findings of the study with any degree of caution given their very limited representation of processes. There is undoubtedly an interesting story from Glendhu, but it cannot be reported in the way of this manuscript. It is too emphatic, and does not command such direct and forthright conclusions!
I think the author has an interesting contribution to make – there are some interesting sample data from the Glendhu catchment that show the watershed to have some interesting characteristics, albeit from limited data. It is entirely reasonable to present this, and to use some simple analysis to suggest the mechanisms of streamflow generation that occur in this (and possibly other) small catchments. However, the manuscript as it stands is an entirely inappropriate framing of a very limited dataset, from a very small watershed, used to produce a very simple model which is subjected to very little scrutiny within the context of alternative approaches.
For these reasons, I do not think this manuscript should be accepted for publication in HESS. This is because there is no scientific argument elucidated to justify the utility of the BRM method over existing (and generally-used) approaches.