This paper describes an experimental work aimed at studying the occurrence of thermal siphons in Rotsee, a shallow lake sheltered by the wind. “Thermal syphon” indicates a physical process driven by differential cooling mainly due to bathymetry, which has important ecological implications, enhancing the hydraulic exchange between the littoral and the pelagic zone. The Authors made use of 1-year velocity and temperature data in the shallower area to detach thermal syphons. They focused on the frequency of occurrence over the year and analysed the forcing data suitable to explain this seasonality.  

They developed a state-of-art experimental work, winding the situ descriptions of the phenomena, which are not so frequent in the literature, especially when aimed at investigating the process over the seasons.

The main limits of this contribution are the weak readability of the paper and the case-specific algorithm proposed to analyse the phenomena. With regard to the first aspect, my suggestion is to be much more concise in the text and in the figures, limiting the number of information to the most relevant ones, or alternative to help the reader to distinguish between the more and less relevant.

More specific suggestions are listed in the followings.

Methods 2.1. The computation of the wind speed doesn’t seem satisfactory for two reasons: 1- the location of the meteorological station is hardly well representative of the wind conditions over the lake’s surface 2 – the methodology to derive wind data from the Lucerne station is not properly justified. Given the sheltering of this lake, a correlation between these sites’ data is unlike and I am quite doubtful about the suitability of a neural network algorithm to estimate it, given the local character on the wind field. Despite I don’t know the typical wind speeds at these site, I believe that in relative terms an error E_RMS  of 0.67 m/s is high. On the contrary, the approach is valuable for the other variables. I think that these sources of uncertainty must be accounted and discussed. Actually I do not have any suggestion regarding the solution of problem 1, apart for discussing this limitation in case of absence of other suitable data. With regard to problem 2, instead, I believe that is necessary to introduce an uncertainty in the fluxes evaluation.  

Methods 2.2.

L 160-161. Is the SW measured or parametrized? From section 2.1 it seems that it is measures but then in 2.2 it seems to be parametrized. Being a widely available parameter, I do not see the need to compute it.

L172. Are there measurements to support the S value? Did you perform any sensitivity to assure that possible variations had no effect on your evaluations at a seasonal scale?

L176. From this paragraph it seems that HQ0 is always a negative loose term, while the shortwave is the only one term that contributes to heating. On the contrary LWin and HC can be positive too. In general the way to manage the signs of the fluxes in this section terms is a bit confusing. I suggest to reason in term of H0net and BOnet only (1+2 eq), without distinguishing between SW and other terms. What is important to verify is whether the net flux is positive or negative. This suggestion should be extended to the other sections of the paper.

Methods 2.4. In the methods aimed at detaching the thermal syphons I would have expected to see the vertical component of the velocity as a target variable, in particular before the beginning of the event. Is there any reasons why you did not mention it? Given the uncertainty on the wind data, I think it could be a better way to distinguish between thermal syphons and wind driven flow (see e.g. Fer et al. 2002).

Table1. Separate the extremes of the range with a “-“ in place of a “,”. Why no range for Ux and Uz?

Figure 3,4,7,8. The figures of this paper are too much dense of information. The effort to make them fully informative has the counter-effect to confuse the reader with too much data and is not efficient in highlighting a clear message. Make an effort to make the figures clearer, with less but more direct information, and eventually reduce the number of figures. ( Fig. 8 in particular is really hard to follow)

Fig. 3. The first panel is useless. In the second panel limit the plot of B0net. Velocity and temperature contours together limit the readability. Look at Fig. 3 of Fer et al. 2002 as an example of a good representation of a single event: a single line of Ux and Uz is much clearer. Finally the contours of temperature between 13.00 and 17:00 looks like affected by an error in the interpolation. Do you have enough thermistors? If so, how do you explain the different pattern? How do you explain the rapid changes in signs of Uz? Please comment.

Results 3.4. The R2 values are really low, I would limit the analysis only to the variables which show at least a trend (not the case of tauf for example). In the conclusion you defined “robust” these relationships , but these R2 do not support these conclusions. I would be more cautious to base the conclusions on the basis of these results.

References. A careful review of the references is needed (for example Rao and Schwab, Meyers and Dale are not present)

HESS three Prinicipal Review Criteria

1. Scientific significance is excellent with new concepts, methods, and data.

2. Scientific quality is good. The approach and applied methods are valid and the results are appropriately discussed.

3. Presentations quality is good. There is an appropriate number and quality of figures/tables, appropriate use of English language. I found that there were few typos and that the English is good. The use of some symbols and abbreviations is confusing.



General Comments



The preprint addresses an important aspect of differential cooling with new concepts and data. With an extensive data set the thermal siphon process is shown to flush the near shore region. A simple model based on practically measured or available data is used to predict this process and it's seasonal variability. I enjoyed reading all five sections and the appendices and found all of the figures engaging. In summary I beleive the work represents a significant contribution to the field and is well suited to HESS. My only general comment is that the clarity of the paper's main findings are obscured somewhat by the complex collection of abbreviations and symbols. In the first three specific comments below I address this and other clarity issues that I think should be addressed.



Specific comments



Three specific comments related to the overall clarity:

1. The along-x locations and their labels are confusing even after the reader is comfortable with the XZ description of the lake:



L_lit is the distance from one end of the lake along the thalweg to MT. So MT was located at the location where the photic zone reaches the bottom? I don't think this is ever stated, rather it seems MT is located at an arbitrarily shallow location along the thalweg.



L_ML the distance from the same origin along the thalweg to the isobath that matches the depth of the mixed at MB.



x_q the distance from the same origin to MT (where q is measured/predicted but not clearly linked between Figures 1 and 2).



l_p the length of the plateau (not indicated in Figure 2).



I think the formatting of these labels should be more consistent (e.g. a capital letter L followed by a subscript) and that x_q or L_lit be omitted. A similar simplification would help with the depths (d_p,d_MT,h_TS,h_lit, etc). I never could figure out what MT and MB stood for.





2. Although the transect data in the schematic represents an efficient use of space and looks great I think it unnecessarily complicates the schematic. The schematic should address the seasonal cycle, identify the plateau, perhaps include the equation q=c_q h (BL)^1/3 or similar equation for U, and  serve as a road map or foreshadowing for the rest of the paper. Something like Table 2 added to the introduction could compliment the schematic. Where is the origin x=0 on the map in Figure 1? Why not identify the plateau in the schematic? Could the authors incorporate a graphic illustrating the essential time scales? If aspects at the end of the paper are too complicated to include in the initial schematic provide a revised schematic at the end of the results or in the discussion. I recognize the authors have spent some time linking the text and figures including Figure 1 and Table 1 together but it still needs improvement.



3. The language related to flow direction is sometimes confusing. I think this is partially due to the fact that the shorelines to the northwest and southeast are closer to both stations than the shore line to the northeast. The authors should explicitly state early in the paper and repeat in several captions that offshore flow is southwestward flow or something similar, line 191 is inadequate. I don't think the authors ever comment on along shore flow, tell the reader why it's ignored or if there's none.



Other specific comments



4. I was expecting to see more transects demonstrating the TS during other times of the year e.g. TS in July, October and December, were there no others collected?



5. Figure 2. Provide the depth at MT in Figure 2 (b). I think the map of Switzerland should be idienfied as a map of Switzerland.



6. Lines 145 to 152 - has anyone ever done this before for winds or humidity? explain why you think the simpler approach failed. Can you provide a separate R^2 for the northerly and westerly wind components, or the along and across axis wind components?



7. Line 239 I think this is ok for B_0 and is discussed later but I'm not so sure about L_ML, wouldn't this often increase over the cooling period?



8. Table 1 would benefit from some recomposition to aid in connecting the four columns, particularly the fourth column, e.g. swap the third and fourth column and justify the 'definition and equation' column left.



Technical corrections



1. Whether limnology is patriarchal or not the reference to 'fathers of limnology' reads a little too patriarchal.



2. The whole sentence beginning 'Such shift' on line 594 needs improvement, to start, change 'Such shift' to 'Such a shift'.



3. line 373 and 374 change shadow to shading.



4. line 376 refer to figure 7 for the histograms.



5. line 475 remind the reader what the depth is at MT.

In this manuscript, the authors use a unique year long time series to estimate the frequency and strength of the thermal siphon in lakes and the influence on flushing of the littoral zone. This is an interesting dataset and addresses an important concept in physical limnology. The writing and presentation are mostly very good. I have a few general comments that I would like to see addressed though:

1. What is the magnitude of the outflow near the study site? Is it important relative to magnitude of the flushing rates?

2. What influence does the three-dimensionality of the littoral zone play? From my understanding, the entire framework here is 2D, but how uniform do you suppose q is across the lake? What are the limits of your results for other lakes in that context? It appears that Rotsee is about 2D as it gets, but is there a littoral zone aspect ratio where this all falls apart?

3. Given the time of year where TS is most prevalent, does this play a role in accelerating autumn turnover?

4. TS is stronger in the summer, but less frequent. As noted, the summer is when this physical process might have the most impact on ecological and biogeochemical processes. Is there a way to compare TS across the months in a more quantitative fashion, for example, the total volume of water flushed in each month?

Figure 3d - the contours on the upper right corner look more like an artefact of the contouring than anything that might possibly be real?

Figure 6 - are you forcing the intercept here? From the equations that seems the case (an intercept of 0), but that doesn’t look like the best fit line, at least for (b)