I reviewed the initial submission of this manuscript. I appreciate the detailed responses to the reviewer comments and actually found many of the responses around model decisions and approaches engaging and insightful. I read the responses prior to reviewing the new version and I was excited to read how the authors had incorporated this material into the revision. Personally I find the model evaluation steps and decisions much more interesting than the climate change simulations. I think we learn more about our systems through testing models and understanding why and where they succeed and fail. In contrast, climate change simulations really amount to a glorified sensitivity analyses that often doesn't provide many new insights into system behaviour or advancing our scientific knowledge. Based on the reply to comments, I was expecting to see more of the 'learning from models' piece in the manuscript. I think some of that discussion is there, but unfortunately, I continue to find the new version of the manuscript difficult to follow and overwhelming, which ultimately reduces the impact of this study. I provide some suggestions on how the presentation could be improved. I also have a remaining concern about some of the arguments around the model over-estimation for summer temperatures for the Alpine catchments. I detail these comments below, followed by some more specific comments.
1. Organization and focus of the paper
I continue to struggle getting through this manuscript and I've spent a bit of time reflecting on why that is. There is an immense amount of material presented which partly contributes to the confusion, but in thinking about this more, I suspect some serious efforts in restructuring and adding focus to the manuscript will ultimately make it easier to read and communicate the key points of this study. Here are a few suggestions:
- Some of the result sections contain methodological details, as well as discussion points. This makes it difficult to follow some of the lines of reasoning, as well as getting the key points from the study. In addition, the reader is constantly referred to different sections ('see Section X', 'presented below', 'details below', etc.), as well as being referred to the (extensive) supplemental material and details in other papers -- which makes it all challenging to follow, even on the third read through, let alone the first. I think by keeping introduction, methodology, results and discussion material in their own sections would improve the logical flow of the paper. I also think the authors should spend some time creating a logical narrative throughout the manuscript that is sufficiently supported with standalone evidence and explanations without having the reader search all over for information. However, you don't want to include so much detail that it results in an overly long paper. I recognize this is a difficult balance to strike with the amount of modelling work done here, but the current version remains difficult to follow and digest.
- The introduction could be substantially streamlined. It jumps around and lacks a logical flow. In addition, the study would be better contextualized with clear statements of the study objectives. Currently, the closest I could find to stated research objectives in the last paragraph of the introduction amount to (a) a statement about which models were employed in this study and (b) that these models were used to investigate the impact of climate change. More specific and testable objectives (or even hypotheses) would give the study greater focus and impact.
- Some detailed editorial work by the authors is needed. The text throughout can be streamlined - such as removing sentences like Page 1 Line 3-4 and Page 1 Lines 8-9 ('This represents...'). Another example is that the paragraph on page 2 lines 16-21 could be removed or incorporated more succinctly into the opening paragraph. These are just examples and I encourage the authors to critically review the writing throughout to cut unnecessary words and sentences. In addition, the the grammar could also be improved throughout.
- A simple suggestion, but perhaps add a prefix to the catchment names to designate whether they are plateau or alpine sites (e.g., P-Suze, A-Lonza or something similar). Keeping track of these throughout the manuscript is difficult, particularly since they are not always grouped together in a logical way (especially in the supplemental document).
2. Rationale for the summer over-prediction
I struggle with the logic around continuing with the climate change simulations despite the over-prediction of summer stream temperature at the Alpine catchments. It seems that the authors are arguing that the missing cooling mechanism is conclusively known (cold water advection not captured by the HSPF scheme), hence why the model over-predicts - but that's okay, since that won't be an important process under future climate conditions. This seems like two big assumptions (the cold water advection term is the missing component and that it won't be important in the future)! Despite the multitude of graphs, I don't think there is one that shows future absolute SWE simulations (although there are percent change figures). Is some SWE simulated for future scenarios? If so, I would argue that even if the cold water advection is the missing term, it will remain a key missing process in these models that is potentially important for correctly simulating river temperatures in these systems; therefore, how much can we believe these predictions?
I appreciate the systematic evaluation of the potential mechanisms that could be accounting for the over-prediction; however, without a way to check some of the internal processes (e.g., evaluating the estimates of incoming solar radiation reaching the stream surface, above-stream wind speeds and humidity, lateral inflow temperatures and magnitudes, potential influences of groundwater/hyporheic exchange), it seems very difficult to conclusively say that the limitation in the HSPF scheme is the single cause driving the model error. What about evaporative cooling - is the latent heat flux underestimated in summer (I believe the alpine rivers are more open and perhaps characterized by greater ventilation - perhaps this is not properly captured in the model)? What about errors in input data? For example, Figures S36-S46 highlight some interesting differences between the measured and CC scenario precipitation estimates. It seems for many of the streams that exhibit over-predictions in summer stream temperature are also associated with under-estimates of summer precipitation (and discharge), such as Lutschine (Figure S44). Is precipitation under-estimated in these alpine catchments (which is often an issue in high elevation catchments) driving the over-prediction of river temperature? Overall, this seems to be a major challenge and undermines the confidence in the climate change scenarios. Similar to my point in the above, I think focusing on this model failure is far more interesting than whether the model simulates X degrees celsius of warming in 50 years, so I don't see this model limitation as barrier to publication. Instead, a repositioning of the manuscript to focus on these model failures and expanding the possible causes of these over-predictions (beyond what is already done), could highlight some important areas of future study to improve our simulations of stream temperature in these types of environments. I realize these types of climate change impact simulations are published all the time, so I'll understand if the authors would like to stick with that focus of the manuscript, but personally I think focusing more on exploring the model failure could make for a more satisfying study.
Some specific comments:
Page 1 Line 8: This sentence could be removed ('This represents...').
Page 2 Line 2: I don't think this sentence is needed here.
Page 2 Line 9: Do you mean 'along with' instead of 'along to'?
Page 2 Line 22-27: The logical development of this paragraph is confusing. It starts with a statement about river temperatures impacting groundwater temperatures, but then jumps into how precipitation patterns will impact groundwater thermal regimes and then concludes with this being a rationale for studying future changes in stream temperature. Some of these points are important, but how they are developed and linked could improved.
Page 3 Lines 1-2: Could you briefly expand on what differences were found in that study instead of just stating there were differences?
Page 3 Lines 7-8: This doesn't seem like the most appropriate reference here as the Horton et al preprint hardly talks about river temperature models.
Page 4 Line 6: Could you explain to non-Swiss readers why having both Swiss Plateau and Swiss Alps catchments is important?
Page 5 Lines 3-4: Do you mean the results of Epting et al 2021 were used in the current study or that the current study results were used in Epting et al 2021?
Page 12 Line 15: 'first' instead of 'fist'
Page 13 Figure 3: Spelling mistake in the legend.
Page 15 Lines 3-4: Are these sorts of influences (e.g., cement factory effluent) common near the observation points in these catchments? I thought these catchments were selected because they had minimal anthropogenic influences?
Page 27 Lines 13-14: The Du et al study was conducted in Canada, not the USA.
Page 28 Lines 16-17: But on Page 4, Lines 8-11 it was stated that these study catchments were selected based on criteria (e.g., no large lakes, no anthropogenic disturbances, no dams) that other Swiss catchments didn't meet; therefore, are they actually representative, since it's well established that lakes, dams and anthropogenic disturbance can have profound influence on river thermal regimes?
Figures S52-S58: Are these daily or hourly values presented in the graphs? If hourly, the upper range in incoming solar radiation seems low around the summer solstice. |
