HESS_2021_158: Author responses to RC-1

 Summary

The authors present a modelling study analysing the effect of forest regeneration on blue and green water fluxes for a catchment in the Scottish Highlands, which have undergone dramatic decreases of native pinewoods since the 17th century. The authors use the tracer-aided ecohydrological model EcH2O-iso (Kuppel et al., 2018a) to model flux partitioning, water ages and hydrological connectivity under three different conditions (i.e., baseline conditions, thicket forest and old-open forest) representing different stages of natural forest regeneration.

The model results highlight that the thicket forest stage leads to the greatest changes in flux partitioning, water ages and hydrological connectivity especially during low flow, while establishment of old-open forest will likely result in the system returning to similar ecohydrological fluxes as during baseline conditions.

The authors argue that this study demonstrates the importance of considering different stages of regeneration as well as their spatial and temporal impact on ecohydrological partitioning to accurately inform landscape restoration.

Response to Summary: Thank you to the reviewer for taking the time to read our work and provide constructive comments to strengthen the manuscript.

General comments

RC-1.1: The study fits the scope of Hydrology and Earth System Sciences and represents an important contribution to investigating the effect of landscape restoration. The study uses existing concepts and methods, but applies them to different landscape scenarios than previous research. Hence, the paper represents a substantial contribution to scientific progress in this field. The paper is well-written and considers an appropriate amount of related work. The figures and tables are well chosen to support the results and conclusions of the study.

Response to RC-1.1: Thank you for recognising the significance and quality of our work.

I do not have major general comments, but I am missing some more in-depth discussion as to (1) the added value of the isotope module and (2) the likelihood of the two land-cover change scenarios under climate change.

RC-1.2: For (1), the authors refer to the validation by Kuppel et al. (2018a), but it would be useful to discuss in the paper to what extent the isotope data helped constrain model parameters and whether the model parameters sensitive to the isotope data are crucial for this study. In view of the uncertainty bounds in the behavioural solutions and to illustrate the value of the isotope data, the authors might want to include a baseline simulation without isotope data and compare the model uncertainties to those of the tracer-aided simulation.

Response to RC-1.2 & RC-1.16: We respond jointly to this comment and RC-1.16 given that they are related. Thank you for the suggestion of running a baseline simulation without isotope data; however, we would argue that this is not within the scope of this paper. We feel that the use of isotopes to constrain storage and mixing volumes in hydrological models to improve confidence in process realism is already well established. Indeed, previous work using simpler models in the Bruntland Burn catchment has shown this to be the case (e.g. Birkel et al. 2011). More recent work has also shown the value of isotopes in better constraining internal process representation and water partitioning in more complex process-based models, such as EcH₂O-iso (e.g. Holmes et al., 2020; Smith et al., 2021). In revision, we will make the established value of incorporating isotopes in modelling clearer in the Introduction (e.g. page 3 L84-85).

The width of the uncertainty bounds in this study are likely symptomatic of the wider and well-documented issue that models such as EcH₂O-iso have many degrees of freedom both in the form of free parameters and spatial patterns of simulated fluxes. Recent work has shown that whilst isotopes can improve simulation of general catchment functioning, they may not be sufficient to fully constrain the detail of all individual processes operating within a system, leading to a persistence of uncertainty in model outputs (e.g. Holmes et al. 2020). This can be further compounded by scaling issues (e.g. Smith et al., 2021) and the extent to which isotopes can constrain GW fluxes >5yrs (e.g. Stewart et al., 2010). The skill of EcH₂O-iso at simulating a range of ecohydrological and isotope datasets in this work increases confidence in its ability to provide plausible realisations of catchment functioning to the extent we can say something useful regarding important issues such as landscape restoration. However, it may also be the case that these realisations could be refined further with additional data. What sort of data would have most value in this regard is an open research question and one that that the authors intend to contribute to in future work; however, we again feel that any significant discussion of this is beyond the scope of this paper. In revision, we will try to communicate this sentiment more explicitly and briefly comment on some of these wider issues in the context of needs for future modelling studies of land-cover change (e.g. Section 5.1 P. 26-27 L. 477-483; Section 6).

RC-1.3: Related to this is also the discussion of changes in water ages with progressing regeneration (section 5.2), which should underline more why this information is highly beneficial for assessing regeneration changes as opposed to looking at the changes in blue and green fluxes only (and thus why we need the isotope data).

Response to RC-1.3: The age of fluxes and their associated storages, along with how these change in response to regeneration, can enhance our understanding of the spatial and temporal aspects of catchment storage-flux interactions and their sensitivity to change. Thus, this is a key advantage of using tracer-aided ecohydrological models and we will highlight this point more clearly on revision.   

RC-1.4: Regarding (2), given that the full regeneration to old-open forest might take several decades, I am wondering whether changing climate might lead to a different trajectory of change than the one depicted in the study. More specifically, how realistic is it that the system can meet increased evaporative demand during summer (e.g., Werritty and Sugden, 2013)? Would it be possible to test this for the study catchment with the EcH2O-iso model (see page 30, lines 595–599)?

Response to RC-1.4 & RC-1.15: We respond jointly to this comment and RC-1.15 given that they are related. Thank you for this comment; we agree that the discussion surrounding the likelihood of the simulated scenarios could be improved. It is difficult to say how climate change may affect the development of the old forest given the large degree of uncertainty in how rainfall distributions will change and, consequently, interact with increased evaporative demands in summer. Exploring this question would be possible with EcH₂O-iso if vegetation growth dynamics were switched on; however, in this application they were switched off so as to minimise the number of processes requiring constraint when only simulating “snapshots” of regeneration. To address this comment in revision, we will add a short final section to the Discussion to cover scenario uncertainty. Along with the current final paragraph of Section 5.1, this will include a brief discussion regarding how details of the old forest structure may be uncertain due to climate change and/or different possible trajectories of regeneration.

RC-1.5: I also have a comment on the data availability. According to the HESS data policy, “data and other information underpinning the research findings are "findable, accessible, interoperable, and reusable" (FAIR) not only for humans but also for machines”. If the data cannot be made publicly available, there should be “a detailed explanation of why this is the case”. Please provide the data in an open repository or explain why this would not be possible.

Response to RC-1.5: Data will be made publicly available via an institutional repository in revision.

Specific comments

RC-1.6: Page 6, line 150: do you mean that there is an exponential decrease of roots in each layer with depth? Please clarify.

Response to RC-1.6: This will be changed to “The fraction of roots in each layer is determined by an exponential function describing how root fraction decreases with depth.”

RC-1.7: Page 6, line 160: could you briefly comment on the impact of this assumption of complete mixing? With a total soil depth of around 30 m in some simulations, how does this assumption affect the water age simulation? I could imagine that the L3 soil layer might contain a relevant proportion of older water, which might bias the water age of transpiration towards older ages using the complete mixing assumption.

Response to RC-1.7: Given the wet, low energy environment of the BB, the complete mixing assumption is likely a reasonable approximation under most conditions given the model grid size and daily time steps. Previous empirical work in the catchment has shown limited evidence of ecohydrological separation beyond minor evaporative enrichment of increasingly mobile waters in the upper soil (e.g. Geris et al., 2015); consequently, the main limitation of the complete mixing assumption is likely to be over-enrichment of water in L1 which may be translated throughout the soil profile (Kuppel et al., 2018) rather than adverse effects on transpiration ages. The effect of L3 on the latter will further be mediated to some extent by the more limited presence of roots in this layer (most vegetation types have roots within 20-50 cm of the surface, most of which will often fall in L1 and L2).

RC-1.8: Page 6, line 168: “soil types were assumed to be spatially uniform”. I am not sure I understand. Do you mean there is only one soil type per cell (as in Fig. 1a) or what exactly is spatially uniform? Also it is not clear to me how to read Table 1: should the percentages across all vegetation types (including bare soil) for each soil type add up to 100%? Could you explain this in a bit more detail in this paragraph?

Response to RC-1.8: By this we meant that the properties of each individual soil type are uniform in space. We will update to: “The properties of each soil type were assumed to be spatially uniform”. It is correct that % cover of all vegetation types including bare soil should sum to 100% for a given soil type. To improve clarity, Table 1 will be reorganised by soil type rather than vegetation type, and it will be noted that vegetation fractions sum to 100% on Page 6 L176.

RC-1.9: Page 9, line 214: how many simulations meet the criterion of simulated saturation areas < 60%? Why are only 30 runs of those retained as behavioural results? This is probably a small proportion of the first subset, but it still gives large uncertainty bounds, for example, in flux ages.

Response to RC-1.9: Approximately 11,000 simulations met the criteria of saturation area <60%. However, many of these simulations will have performed poorly with respect to the other calibration targets used as part of the multi-criteria approach. We retained 30 behavioural runs to strike a balance between the need to illustrate uncertainty in model outputs and the increased computational demand of running the model when producing spatial outputs required for the change analysis but not calibration. We will provide this justification on Page 9 L 214. Also see Response to RC1.2 & 1.16 for comments on the width of uncertainty bounds.

RC-1.10: Page 12, lines 282–283: I do not fully understand. What kind of threshold and what is the role of reinfiltration along a flow path? Please clarify.

Response to RC-1.10: In some connectivity analysis (e.g. Turnbull and Wainwright, 2019) a threshold of simulated overland flow is used to determine if a cell would be measurably connected in reality. In EcH₂O-iso, overland flow is accumulated along a given flow path but with the potential for losses to re-infiltration, potentially preventing connectivity of cells to the stream. Because of this, we did not opt to impose a potentially arbitrary threshold of overland flow for a cell to be considered connected and instead inferred connectivity directly from simulated spatial patterns of overland flow. In revision, we will clarify this on Page 12 L280 with: “In EcH₂O-iso, losses to re-infiltration along a given flow path can prevent upslope cells producing OLF from connecting to the stream (Maneta and Silverman, 2013). Consequently, connectivity was inferred directly from simulated spatial patterns of OLF without imposing a minimum threshold of OLF below which a cell would be considered disconnected (c.f. Turnbull and Wainwright, 2019).”

RC-1.11: Page 13, lines 290–291: could you also state the values of the performance metrics for behavioural runs?

Response to RC-1.11: The values for performance metrics are currently given in Table 2. Given the number of calibration targets, we would not be keen on moving these values into the text to ensure that readability is maintained. 

RC-1.12: Page 27, lines 507–510: “Greater consistency…”. I am not sure I understand. Do you mean that regeneration does not affect the fluxes during larger events because of sufficient amount of rainfall and stored pre-event water during these events?

Response to RC-1.12: This is correct. The relevant sentence will be changed to “The lesser impact of thicket forest on the simulated magnitude of high flows suggests that increases in storage capacities (Fig. 4) and “green” water fluxes (Table 3) were insufficient to moderate the combined influences of antecedent conditions and precipitation inputs that led to the largest events modelled here (Fig. 5b).”

RC-1.13: Page 28, lines 517–520: So, would that mean that the old forest state might be achieved much later or maybe not at all?

Response to RC-1.13: In this case, the old forest state would still be reached, however the configuration may be different to the one simulated here. In particular, whilst old forest would still be present on the hillslopes, it could be that drying out of the valley bottom results in the presence of a younger regenerating forest rather than persistence of bog woodland/vegetation. This will be clarified in the new final Discussion section highlighted in Response to RC-1.4.

RC-1.14: Page 29, lines 554–555: I do not see big differences in the connectivity changes between low / moderate summer events and the large winter event. Could you support this assertion by mentioning percentage changes in section 4.7?

Response to RC-1.14: Thank you for this comment. On reflection, our choice of events presented for the connectivity analysis did not best support our assertions regarding the effect of regeneration on connectivity as the August 2014 event was actually quite large with relatively wet antecedent conditions. In re-evaluating this part of the analysis, we have found we can add greater strength and nuance to our assertions regarding the effects of regeneration. This will result in the following changes:

• Section 4.3: Highlight that, proportionally (as revealed by plotting ln(Q) in Fig 5b), decreases are greatest for low to moderate flows in late summer and during autumn/winter rewetting, whilst large winter events can somewhat “reset” the catchment towards baseline conditions, resulting in more limited divergence of subsequent winter/spring flows.
• Figure 9: Include a small/moderate event in late summer or during autumn/winter rewetting; with the drought event, this will help strengthen the point that decreases in connectivity are greater under these conditions rather than larger events in summer and, particularly, winter. In response to Reviewer 2, the figure will also be condensed by removing the histograms from this plot and instead plotting the spatial distribution of cells connected in at least 50% of behavioural simulations, coloured by their flow path lengths.
• Section 4.7: Update to accommodate new figure and include % changes in cell connectivity.
• Section 5.1, Paragraph 3: Refine text to indicate that storage deficits from reduced GW recharge and increased summer transpiration not only affect late summer baseflows but also delay rewetting and affect the magnitude of small/moderate events at this time.
• Section 5.3: More nuanced discussion of connectivity changes for low/moderate summer and rewetting events vs. larger events in summer and winter.

RC-1.15: Section 5.3: see general comment (2): I would appreciate some words on the likelihood that regeneration would undergo these two forest stages in view of climate change. Could it be that less rainfall/higher ET in summer would lead to a diversion of pinewood regeneration as depicted in Fig. S1 such that increased transpiration demand of thicket forest could not be met and transition to old forest would not occur? This links to the statement on exploring trajectories of change made in the Conclusions.

Response to RC-1.15: Please see response above to RC-1.4.

RC-1.16: Page 30, line 590: see general comment (1): I am not sure about the benefits of the isotope observations here. Do we need the isotope module of the model or what additional validation data might be useful to constrain the uncertainty bounds? Could the authors comment on the uncertainty that would result from calibration without isotope data? Is it the comparably low temporal and spatial resolution of soil-water isotope data that limits the uncertainty reduction?

Response to RC-1.16: Please see response above to RC-1.2.

Technical corrections

RC-1.17: Page 5, line 142: gridded

Response to RC-1.17: This will be corrected.

Tables

RC-1.18: Table 1: how did the authors determine the exact proportional aerial coverage in the two scenarios? References are given here but it is not clear to me whether/how these numbers have been derived from the information provided in the references

Response to RC-1.18: Field descriptions from Steven and Carlisle (1959) was used to determine the understory composition of bog woodland. The sum of understory vegetation % cover in Table 1 of Parlane et al. (2006) was used to determine the fraction of heather under thicket and old-open woodland. Bog pine % cover was based on figures for uncut bog in McHaffie et al. (2002; p.214) and plan drawings from Summers (2018). Percentage covers of thicket and old-open pine were derived from plan drawings of relevant stand canopies in Figure 3 of Summers et al. (1997). These details will be briefly added to the notes of Table 1.

Figures

RC-1.19 Figure 1a: would it make sense to have more meaningful symbols and/or colours for the monitoring sites, grouping weather stations, soil types and vegetation?

Response to RC-1.19: We will update the symbols/colours of Figure 1a to better group monitoring sites.

RC-1.20: Figure 1: could you also include a digital elevation model, so it is easier to see the location of the hillslopes in the catchment?

Response to RC-1.20: We will add contour lines to Figure 1a to improve visibility of the hillslopes.

RC-1.21: Figure 9: could you also show the dates of the different snap shots directly in the figure? If not, the reader has to switch back and forth between the figure panels and the figure caption.

Response to RC-1.21: Event dates will be added to Figure 9.

………………………………………………………

New references not currently in manuscript

Birkel, C., Soulsby, C., Tetzlaff, D., (2011). Modelling catchment-scale water storage dynamics: reconciling dynamic storage with tracer-inferred passive storage. Hydrological Processes 25: 3924-393. DOI: 10.1002/hyp.8201

Geris, J., Tetzlaff, D., McDonnell, J., Anderson, J., Paton, G., Soulsby, C., 2015. Ecohydrological separation in wet, low energy northern environments? A preliminary assessment using different soil water extraction techniques. Hydrological Processes 29: 5139-5152. DOI: 10.1002/hyp.10603

Holmes, T., Stadnyk, T., Kim, S.J., Asadzadeh, M., 2020. Regional calibration with isotope tracers using a spatially distributed model: A comparison of methods. Water Resources Research 56: e2020WR027447. DOI: 10.1029/2020WR027447

Smith, A., Tetzlaff, D., Kleine, L., Maneta, M., Souslby, C., 2021. Quantifying the effects of land use and model scale on water partitioning and water ages using tracer-aided ecohydrological models. Hydrology and Earth Systems Science 25: 2239-2259. DOI: 10.5194/hess-25-2239-2021

Stewart, M. K., Morgenstern, U., McDonnell, J. J., 2010. Truncation of stream residence time: How the use of stable isotopes has skewed our concept of streamwater age and origin. Hydrological Processes 24: 1646–1659. DOI: 10.1002/hyp.7576

Yang, X., Tetzlaff, D., Soulsby, C., Smith, A., Borchardt, D., 2021. Catchment functioning under prolonged drought stress: Tracer-aided ecohydrological modelling in an intensively manged agricultural catchment. Water Resources Research 57: e2020WR029094. DOI: 10.1029/2020WR029094

HESS_2021_158: Author responses to RC-2

Summary

The manuscript by Neill et al. presents an ecohydrological modelling study about structural changes of forest regeneration and the effect on water flux portioning, water ages and hydrological connectivity. They use the EcH₂O-iso for a small experimental catchment in the Scottish Highlands and simulate a baseline and two land cover change scenarios, a thicket and an old-open forest.

The modelling study gives the opportunity to create an old-open forest which might be very difficult to create in a field experiment due to long time period over 100 years, and agriculture forest use (tree age around 40 years). This stage of old forest might happen if the forest harvesting stops, hence especially for stakeholders it is interesting to see the influence of such forest development. But also, the research community gets an idea about the effects of a thicket and an old-open forest to the hydrological conditions. This could help to see the field experiments with a different angle and to support information around such experimental sides.

The text is well structured with meaningful subheadings and well-structured paragraphs. The manuscript is in the scope of the HESS journal and gives new insights in the field of tracer-aided ecohydrological modeling.

I see an especially need to strengthen the text for an easier readability with less abbreviations and clear sentences. The figures and table also need some revisions for an easier readability, e.g. bigger fonts. Here I give some general comments and specific comments at the end. (“Line” is abbreviated with “L”).

Response to Summary: We appreciate the careful review and positive comments.  We are also grateful for the many constructive suggestions provided by the reviewer, although note that in a lot of cases these relate to very minor details of the manuscript and its presentation. We therefore hope the reviewer can appreciate that we do not always find the need to accommodate all their suggestions. Please find our individual responses below.

General comments

Abbreviations:

RC-2.1: I suggest to reduce the abbreviations for easier and an undisturbed readability. Especially since some abbreviations are just used a few times (e.g. SW 4x, VWC – 6x, RZ – 7x, OLF-13x). From my point of view, I would only keep LAI and use the full words for the others. ET and GW, might be an option to keep as well, but it still interrupts the reading (as an alternative, a table with all the abbreviations could also work).

Response to RC-2.1: We will use the full words for abbreviations that appear fewer than 10 times to improve readability. Respectfully, however, we believe the use of abbreviations occurring more frequently is justified given the manuscript length, as is use of the common abbreviations ET and GW.

RC-2.2: Some abbreviations are not introduced in the text e.g. NE (Line 286), SE (Line 463), NW (Tab1, L 385, L403)

Response to RC-2.2: As these refer to commonly used ordinal directions, we did not see the need to define. However, to improve readability these directions will be written out in full.

RC-2.3: Leaf area index is mentioned in L 93, but introduced in L 186, I would also suggest not to introduce the LAI in italic, or if this is really necessary only use the italic version, also for figures and tables

Response to RC-2.3: LAI will be introduced in L93 instead. When LAI is in italics, it indicates that it is the model parameter being referred to; however, we now see that this is not necessarily obvious and will clarify in L186.

RC-2.4: Bruntland Burn, I would suggest to keep the whole name instead of just BB

Response to RC-2.4: Respectfully, we would choose to keep the abbreviation BB when referring to the catchment as this has been common practice in previous papers based at this site.

Figures:

RC-2.5: In general, the figure captions are quite short, maybe some more information for the reader to understand the meaning of each figure can be added.

Response to RC-2.5: Thank you for this comment, we will add more information as appropriate when responding to the specific comments for each Figure.

Fig. 1:

RC-2.6: The symbols should be the same for the same type of station e.g. gauging station in the river (one symbol), weather station (another symbol), ect.

Response to RC-2.6: We will update the symbols in Figure 1a accordingly.

RC-2.7: DW abbreviations should be explained (unclear for me what it could be)

Response to RC-2.7: DW refers to deeper groundwater well. This will be defined in the figure caption.

RC-2.8: Map in the left corner is too small, better to use a bigger map with some parts of Europe to show persons from everywhere, where the catchment is located

Response to RC-2.8: We will make the context map of Scotland bigger. Respectively, however, we disagree that a more expansive map including Europe is necessary as this will likely make it more difficult to see exactly where in Scotland the catchment is located.

RC-2.9: The whole figure looks a bit unstructured and a bit chaotic, maybe it is better to split in two figures

Response to RC-2.9: Following the reviewer’s later suggestion, we will move c-g) to Figure 2.

RC-2.10: 1g) Bog pine, it seems that there is no bog pine at all, is that right? Order the scaling of the vegetation fraction to undifferentiated

Response to RC-2.10: No, this is not correct. There are some cells with a small cover of bog pine, typically less than 10%. To improve clarity, we will replot c-g) so that 0% cover does not show.

RC-2.11: The font size of “Vegetation fraction” including number is too small, also the legend of h)

Response to RC-2.11: We will increase the font size.

RC-2.12: The font size of the headings of c) to g) could be little bit bigger

Response to RC-2.12: We will increase the font size if possible when incorporating c-g) in Figure 2.

RC-2.13: In Tab 2, Groundwater wells are mentioned. Maybe you can also include the location in this figure

Response to RC-2.13: As noted above, groundwater wells are denoted by DW1-4 and will be defined as such in the figure caption.

Fig. 2:

RC-2.14: What is “Regen-baseline”, better (regeneration – baseline scenario), since there is enough space to write the full text

Response to RC-2.14: This will be written out in full.

RC-2.15: Maybe it is an option to include Fig 1 c) to g) in Fig 2, to reduce the overloaded Fig 1.

Response to RC-2.15: Thanks for this suggestion, we will move Fig. 1 c-g) to Fig. 2.

Fig. 3:

RC-2.16: Please add a legend to every subfigure, starting with first observation, second spread, or the other way around.

Response to RC-2.16: Legends will be added to all sub-figures.

RC-2.17: b) (In m³ s-1), guess it is just (m³ s-1)

Response to RC-2.17: No, this is not correct – Ln denotes that it is the natural logarithm of discharge that is plotted, which improves visibility of lower flows. This will be clarified in the figure caption.

RC-2.18: Font size could be a bit bigger, for easier readability

Response to RC-2.18: Unfortunately, font sizes have already been optimised for this figure given the large amount of data to present and, therefore, cannot be made bigger.

RC-2.19: For the caption I would suggest: a) Precipitation; b) and of observed and simulated Discharge; c) […]

Response to RC-2.19: Respectfully, we opt to leave the caption as it is since “observed and simulated” relate to all subsequent variables after precipitation, not just discharge.

Fig. 5:

RC-2.20: Discharge again (In m³ s-1)

Response to RC-2.20: Again, this is correct and will be clarified in the figure caption.

RC-2.21: What is the brown color? The red on top of green? This is hard to see, even for a non-color-blind person (maybe you can find other colors e.g. red and green is not visible for many persons)

Response to RC-2.21: Thank you for this comment. The brown colour is where simulations for the thicket and old-open forest overlap. We will change the colour scheme to be colour-blind friendly and make a note in the figure caption in relation to the colour of overlapping simulations.

RC-2.22: Caption maybe: c) Stream water …

Response to RC-2.22: This will be incorporated.

Fig. 6:

RC-2.23: “Baseline:” It is better mentioned it in the Figure caption, but not as a heading, if it is always the same for all cases.

RC-2.24: I would also suggest to write groundwater instead of GW and evapotranspiration instead of ET, since it is enough space to write the full word.

RC-2.25: Font size should be a bit bigger, for easier readability

Response to RC-2.23 to RC-2-25: Respectfully, we do not see the need to adopt these changes as they are very minor relative to the effort needed for incorporation (especially regarding increasing font sizes given the large size of the figure already).

RC-2.26: For comparison, it would be much easier to read and compare the subfigures, if the “spread”- median daily average would always be the same size. e.g. from 0 to 30 or so for the blue ones and 0 to 2.5 for the green ones.

Response to RC-2.26: Whilst we understand the reviewer’s point here, we are keen to use separate scales for each set of fluxes so that the spatial patterns and differences between seasons for each flux can be clearly identified.  

RC-2.27: What are the brown pixels in a) and e), please explain e.g. in the figure caption

Response to RC-2.27: EcH₂O-iso does not simulate overland flow for cells containing a stream channel nor groundwater outflow for outlet cells. Consequently, there is no value of overland flow or groundwater outflow to show for these cells, so the brown base map is visible. This will be explained in the figure caption.

RC-2.28: Caption: please define the abbreviation “L1”

Response to RC-2.28: This will be replaced with “soil layer 1”.

Fig. 7:

RC-2.29: What are the brown pixels in a) and e), please explain e.g. in the figure caption

Response to RC-2.29: EcH₂O-iso does not simulate overland flow for cells containing a stream channel nor groundwater outflow for outlet cells. Consequently, there is no value of overland flow or groundwater outflow to show for these cells, so the brown base map is visible. This will be explained in the figure caption.

RC-2.30: To get an easier overview I would suggest to write the month in the middle over the first and second subfigure column, and the third and fourth subfigure column, since they are always showing the same time frame, just the scenarios are different.

Response to RC-2.30: Respectfully, we do not feel the effort needed for such a minor change is warranted.

RC-2.31: Again, please use the same spread for all figures maybe 0.5 to -9

Response to RC-2.31: Whilst we understand the reviewer’s point here, we are keen to use separate scales for each set of fluxes so that spatial patterns in the differences of each flux can be clearly identified.  

RC-2.32: Caption: e) GW flow instead of Groundwater flow

Response to RC-2.32: This will be changed.

Fig. 8:

RC-2.33: Again, please use the same spread for all figures maybe +1 to -1

Response to RC-2.33: Whilst we understand the reviewer’s point here, we are keen to use separate scales for each set of fluxes so that spatial patterns in the differences of each flux can be clearly identified.  

RC-2.34: Again, I would suggest to write the month in the middle over the first and second subfigure column, and the third and fourth subfigure column.

RC-2.35: Font size should be a bit bigger, for easier readability

Response to RC-2.34 and RC-2.35: Respectfully, we do not feel the effort needed for such minor changes is warranted.

Fig 9:

RC-2.36: For an easier overview you might consider to include the timing, so 22 July 2013, 10 August 2014 and 30 December 2015 or dry summer period, summer wet period and 100-year return period flood, or something like this

Response to RC-2.36: Thank you for this suggestion, we will add the event dates to the figure itself.

RC-2.37: The figure caption does not fit on the same page, so the figure must be small, but it is no option to just minimize the total figure, since already now the text and numbers are very hard to read

Response to RC-2.37: To condense this figure, we will remove the histograms. Then, instead of plotting the proportion of behavioural runs a cell is connected on the maps, we will plot cells that are connected in at least 50% of behavioural runs and colour them by their flow path lengths. This will allow the figure to convey very similar information but in less space, whilst also allowing modifications to the figure as detailed in the responses to Reviewer 1.

Tables:

RC-2.38: In general, the table captions are quite short, maybe some more information for the reader to understand the meaning of each figure can be added.

Response to RC-2.38: Thank you for this comment, we will add more information as appropriate when responding to the specific comments for each Table.

Tab 1:

RC-2.39: The whole “cover” column should be left-justified, or why is only “As baseline” right-justified?

Response to RC-2.39: “As baseline” is not right justified within the Cover column but instead extends across the Cover, Height and LAI Scale Factor columns as these factors are unchanged in the regeneration scenarios for the relevant vegetation types.

RC-2.40: Instead of just “cover” maybe “proportional aerial coverage” or something like this

Response to RC-2.40: This will be updated in revision.

RC-2.41: The use of italic is confusing, maybe use bold instead of italic for “Baseline”, “Thicket woodland”, …

Response to RC-2.41: Underlining will be used in place of italics.

RC-2.42: Notes a): “pre-existing” with small letter

Response to RC-2.42: This will be corrected.

Tab 2:

RC-2.43: What does “Full” stand for? Full time period? Then maybe also mention again how long this study period is or from x to x.

Response to RC-2.43: Full stands for full study period – we will indicate this for clarity.

RC-2.44: I would suggest to only use “and” or “&”, not both in the same table

Response to RC-2.44: We will update the Table to ensure consistency.

RC-2.45: You might want to explain the A and B behind Forest and Heather

Response to RC-2.45: A and B refer to north- and south-facing sites, respectively; however, we would argue that in the context of this paper, it is sufficient that A and B simply denote different sites.

RC-2.46: Where is the location of “deeper well” 1 to 4, maybe include in Fig 1.

Response to RC-2.46: Please see responses to RC-2.3.

Tab 3:

RC-2.47: Is it really necessary to give the decimal place, full numbers are easier readable (like done in Tab 4)

Response to RC-2.47: We will update the table so that only full numbers are presented.

RC-2.48: What is the added valued to include the second columns with the differences in seasonally averaged flux totals → The table is quite confusing, so maybe it is better shorten the given information, if possible (this also applies for Tab 4)

Response to RC-2.48: The added value of showing the differences in seasonally averaged flux totals is that they more clearly indicate whether the simulated direction of change was consistent amongst behavioural models. This may not be obvious from considering summaries of the seasonally averaged flux totals themselves.

Words:

RC-2.49: Please stick to one version of “old-open” or “old open” including the abstract, tables and figures

Response to RC-2.49: This will be checked for consistency in revision.

RC-2.50: Instead of Oct-Mar and May-Sep I would introduce the words of summer and winter or, dormant season and biological active season, beside a better readability this might also be an improvement of the figures, if you want to stick to the month, I would suggest to write the full names like October to March.

Response to RC-2.50: We will introduce Oct-Mar and Apr-Sep as the dormant and biologically active seasons, respectively, and make appropriate changes throughout the manuscript. 

Introduction:

RC-2.51: L 95 – 99 You explain, that the soil properties are held constant, but then further describe that they might change. I guess, it would be very interesting to see the effect of soil property changes. How strong is the effect here?

Response to RC-2.51: Here, we sought to justify our choice to keep soil properties constant by arguing that changing them may increase uncertainty in model outputs because a) it is not known how any physical changes to soil properties would be expressed in changes to effective model parameters, and b) it is unclear if/how soil properties might change under coniferous forest because there may be processes operating that counteract one another. Consequently, whilst it is desirable to account for changes in soil properties when modelling land cover change, more research is likely needed regarding how exactly properties change under different land cover types and how these changes translate into modifications of effective model parameters. This is indicated in Section 6 (page 30 L591-595). 

RC-2.52: Also, climate change has an important impact to the soil and plants, especially in a 100-year scenario. Maybe you can further explore this part in the introduction or later on.

Response to RC-2.52: Thank you for this comment. To also address comments from Reviewer 1, we will add a short section at the end of the Discussion to comment on possible uncertainty in the modelled scenarios due to factors such as climate change.

3.1 The EcH2O-iso model:

RC-2.53: The model description part with its concept is a bit imprecise.

Response to RC-2.53: We apologise but it is not clear from this comment what exactly the reviewer would like us to change. In revision we will ensure that the description of EcH₂O-iso is as clear as possible.

RC-2.54: The kinematic wave model in the groundwater context (L.154 -160) is not so common, it is normally known for open channel routing. Maybe you can explain this point a bit more detailed. From the description, the term GW is maybe not the right one in L 157, maybe it is interflow? Is there an exchange of river and groundwater (in one or both directions)?

Response to RC-2.54: We will further explain the GW routing mechanism. In EcH₂O-iso, only water in excess of field capacity in layer 3 can move laterally in the sub-surface. This water is conceptualised as GW and therefore this is the appropriate term in L157. There is a one-way exchange between the stream and GW whereby the latter can seep into the former (L159); we will briefly expand on the details of this interaction.

Calibration:

RC-2.55: Give less references to the specific parts of the figures and table. e.g. L339 to 342 (Tab 3) at the end of the sentence is enough. There are many other places where the references to Tables and Figures can be reduced for a much easier readability, without losing information. (e.g. L345, L 346, L 355 (when the whole paragraph is about the figure introduces at the beginning it is not necessary to refer to all the subfigures after each sentence.)

Response to RC-2.55: In revision we will ensure that references to figures and tables are appropriately succinct.

RC-2.56: 4.1 Baseline calibration: refer more to the Table 2, e.g. with the MAE for discharge.

Response to RC-2.56: OK.

Discussion:

RC-2.57: Sometimes difficult to read, especially the very long sentences: L 480 – 483, L 491 – 494, L 507 – 510, L 542 – 545, L 560 – 563

Response to RC-2.57: In revising the manuscript we will endeavour to reduce the length of overly long sentences throughout.

RC-2.58: Here you introduce the terms of dormant season and biological active season (L490 – 491), and winter and summer (e.g. L 505), but without giving the month you refer to in you catchment.

Response to RC-2.58: We will introduce Oct-Mar and Apr-Sep as the dormant and biologically active seasons, respectively, and make appropriate changes throughout the manuscript. 

Specific comments:

RC-2.59: L 39 – 42: very long sentence, please split in two

Response to RC-2.59: This will be changed.

RC-2.60: L 82: maybe delete “which”

Response to RC-2.60: “Which” is necessary here.

RC-2.61: L90: maybe give the catchment area in brackets, and not only call it small

Response to RC-2.61: Respectively, we do not think it necessary to quote the size of the catchment here since it is introduced shortly after in Section 2.

RC-2.62: L94 – 99: changes in soil properties are not included in the model, but here explained that it is very likely to happen. Why are you not including soil property changes when you think they are happening and important? I guess it needs more thoughts why you did not include them. Also, a connection from the missing soil property changes to the specific objectives of the manuscript would be helpful.

Response to RC-2.62: Please see Response to RC-2.52.

RC-2.63: L 106: reference to Fig 1, not only Fig 1a, the whole figure gives information about the catchment

Response to RC-2.63: OK.

RC-2.64: L 116: (SNH, 2016) instead of [SNH, 2016]

Response to RC-2.64: This will be changed.

RC-2.65: L 123: Maybe better: Mean annual precipitation is 1000 mm and potential evapotranspiration is 400 mm, with the […]

Response to RC-2.65: This will be changed as suggested.

RC-2.66: L 125: Maybe better: […] mean temperatures ranging between 1 ℃ in winter and 13 ℃ in summer.

Response to RC-2.66: This will be changed as suggested.

RC-2.67: L 128: please include catchment after BB, also in the other cases in the manuscript so “… BB catchment” e.g. L 197, L 210, …

Response to RC-2.67: This will be incorporated in revision.

RC-2.68: L 149: please explain the soil layer L1, L2 and L3. Is the L1 the top most? How are they defined, maybe with the soil horizons? Or just with a given depth?

Response to RC-2.68: L1 is the top-most layer. The depth of each layer is typically a free parameter requiring calibration (the case in this application). This will be indicated within the description of EcH₂O-iso.

RC-2.69: L 155: please give the source of the Green-Ampt model

Response to RC-2.69: Mein and Larson (1973) will be given as the appropriate reference for the implementation of the Green-Ampt model in EcH₂O-iso.

RC-2.70:L 168: what is meant by “spatially uniform”, please describe further

Response to RC-2.70: By this we meant that the properties of each soil type are uniform in space. We will update to: “The properties of each soil type were assumed to be spatially uniform”.

RC-2.71: L 182: better: 100 m x 100 m grid

Response: to RC-2.71 This will be changed as suggested.

RC-2.72: L 183: add “in the supplementary Table S1.” Or something similar, to know where to find the table, since it is not in the manuscript itself. Also, at other places when referring to the supplementary material e.g. L 185, L 190, L221, …

Response to RC-2.72: References to supplementary material will be clarified throughout the manuscript.

RC-2.73: L 188: What kind of channel? River channel?

Response to RC-2.73: Yes, for the river channel – this will be clarified.

RC-2.74: L 207: “to avoid over-emphasising high flows” – compared to what? Compared to NSE?

Response to RC-2.74: Yes, this is compared to metrics based on mean squared errors, such as NSE. This will be clarified.

RC-2.75: L 271: add … periods of biological growth and dormancy in our study area. Or something similar

Response to RC-2.75: We will introduce Oct-Mar and Apr-Sep as the dormant and biologically active seasons, respectively, and make appropriate changes throughout the manuscript. 

RC-2.76: L 289: model skills instead of model skill

Response to RC-2.76: Skill should be singular here, but we will rewrite “model skill” as “skill of the model” for clarity.

RC-2.77: L 291: Tables 2 and S2, since the supplementary, should just give additional more detail information, so is less important and should be mentioned as a second.

Response to RC-2.77: This will be changed as suggested.

RC-2.78: L 369: “zero” instead of “0”

Response to RC-2.78: This will be changed as suggested.

………………………………………………………

New references not currently in manuscript

Mein, R. G., & Larson, C. L. (1973). Modeling infiltration during a steady rain. Water Resources Research 9(2): 384–394.