Thank you for your thoughtful comments that were echoed by the other referees and will improve the revised manuscript. Please see attached document for the response to all RC.
I agree with reviewer #1 on the high potential of this well conducted study on CH4 emissions from a temperate eutrophic reservoir which includes 2 years of continuous monitoring of total CH4 emissions by eddy covariance (EC) and gap-filling with ANN and ebullition with automated bubble traps at shallow and deep sites and six extensive field surveys during which diffusion (floating chambers) and ebullition (manual bubble traps) were measured at more than 10 sites. The interpretation on the spatial and temporal variability of CH4 emissions can be done on the basis of meteorology (Rainfall, temp, atmospheric pressure), energy balance (H, LE), hydrodynamics (Brunt-vaisala Freq, temp profiles), hydrology (water inputs, water levels) and biogeochemistry (O2, Chloa).
Major comments
My first major comment is about the result section which does not depict the whole dataset. Indeed, only CH4 fluxes are described but not correctly (see below).
Information on meteorology and hydrology would be very welcomed. Description of the energy balance, thermal stratification and its spatial variability, vertical biogeochemical stratification (O2, CH4…) and their spatial variability and chlorophyll a data and its spatial variability are required
For CH4 emissions, I would recommend to separately describe ebullition (funnels, bubble traps), diffusion (floating chambers) and total emissions from EC. As a matter of fact, I wonder whether the gap-filling is not already a kind of interpretation as the gap-filling is based on the covariation of the fluxes with other variables when EC data are available. Therefore, it has to be decided by the authors to keep it in the result section or move it to the discussion. Independently of where the gap-filled fluxes are described (results or interpretation), it would be very informative for the reader to have information on the validated fluxes (“real data”) and on the EC fluxes after gap filling for comparison.
The second major comment is related to the absence of information regarding the calculation of total emissions from the reservoir. A critical discussion on the comparison of the different type of measurements is required in order to determine the adequate methodology to combine them for a robust estimation of total emissions. We currently ignore whether the emission factor given in the manuscript is an average of all measurements, whether it is only based on EC… Did the author take into account the bathymetry for the extrapolation of ebullition from the reservoir since ebullition at deep sites is lower than at shallow sites?
Minor comments
-Throughout the manuscript: Does “Static pressure” depict atmospheric pressure or the sum of atmospheric and hydrostatic pressure?
-Did the author explore the role of hydrostatic pressure (water level and their variations) on CH4 emissions?
-Did the authors attempt to decipher diffusive fluxes and ebullition from the EC dataset (at least when they have concomitant surface concentrations and or chamber measurements with EC measurements)?
As the manuscript require substantial rewriting/reorganization in order to properly present the dataset and better focus on key results in the discussion no detail comments are provided.
Thank you for your thoughtful comments that were echoed by the other referees and will improve the revised manuscript. Please see attached document for the response to all RC.
This paper deals with methane emissions in a small temperate eutrophic lake. Emissions were assessed from a variety of measurement techniques (floating chambers, submerged funnels and eddy covariance) together with some environmental parameters (sediment temperature, atmospheric pressure, heat fluxes, met data…) and a neural network (ANN) approach. The paper discusses the links between CH4 fluxes and the biophysical parameters, as well as it provides an analysis of the temporal and spatial variability of those emissions. The subject is of great interest since methane emissions from reservoirs are still poorly studied and constrained at the global scale. There are very few eddy covariance-based studies with long series (2 years) as presented here. As stated before by reviewers #1 and 2, There is no doubt that the data base gathered here is worth publication in the Biogeosciences journal. Some rearrangements would be welcome before publication.
One of the most striking results presented here is the difference between 2017 and 2018 seasonality and cumulated emissions. Unfortunately, though well argued, there are no direct measurements of nutrients and carbon (TOC, DOC, POC, quality of OM) to support these assumptions. Discussion on the diurnal patterns is also a bit disappointing since the results are not unequivocal.
Authors should focus the paper on the main findings which can be supported by the data provided in the paper, and subsequently, present figures might be a little bit too numerous in that perspective of a more focused paper.
The end of the abstract is mentioning "…there is a trade-off in intensive measurement of one water body versus short-term and/or spatially limited measurements in many water bodies", and also "The insights from multi-year, continuous, spatially extensive studies like this one can be used to inform both the study design and emission upscaling from spatially or temporally limited results". These statements are indeed interesting and I wish the paper would give clearer insights and develop more on this matter in the discussion and conclusion.
Rearrangements suggested by Rev 1 and 2 would improve the paper a lot since results and discussion are all mixed together at the moment. I am particularly sensitive to the place devoted to ANN gap-filling and on the way it impacts final emission numbers.
Minor comments
Page 4, line13: How was used time-lapse camera in this study?
Page 4, line 27: there were no u* filtering at EC-S1? If so, you should argue xwhy
Page 5, line 33: more details are needed on the way Akaike information criterion (AIC) was used to determine fitting rate of change in the chambers.
Page 6, line 10-11: vertical profile were done manually, detail procedure( how long for each level)
Page 6, line 30: give more details about:” a probability design that has been shown to reduce uncertainty relative…”
Page 9, line 26: you should give the information that “both quantitative analyses of the relationship between FCH4 and SedT yielded statistically significant results” before implying a link between those two parameters in lines 22-24
Page 11, line 3: I understand that the sandy substrate mention here was brought for recreation use (beach). Is there any point to measure fluxes at the very specific place?
Page 11, lines 23-24: comment on absolute and relative importance of each factor
Page 11, lines 28, 29 and 30: table 3 instead of table 2
Page 13, line 4-5: any assessment of the mentioned transfer?
Page 13, line 21: any nutrients data to support the suggestion mentioned here?
Page 13, line 26-27: any measurement of residence time and output/input of C to support this?
Page 14, line 2: is this consistent with kinetic found by Grasset et al, 2018?
Page 14, line 28: pattern and patterning instead of patter and pattering
Page 15, line 32: detail input parameter of the model used
Page 15, line 33: Del Sontro et al 2018 ref missing or is this Del Sontro et al 2016?