A long-term dataset of simulated epilimnion and hypolimnion temperatures in 401 lakes (1959-2020)

The manuscript presents a dataset of simulated lake water temperatures of 401 French lakes using a modelling approach. The modelling approach is well explained as well as the choices of how to calibrate (or not) the model parameters. I appreciate the fact that authors discuss the limitations of the dataset as well as the suitable applications, making clear for which purposes this dataset would be limiting.

I think that the presented dataset is relevant to leverage lake research in the context of climate change. The manuscript is overall well written but I suggest some modifications that I list in the attached pdf.

In this study, Sharaf and co-authors present a regional long-term water temperature dataset (LakeTSim: Lake Temperature Simulations) for 401 French lakes by combining numerical modelling and satellite thermal data. The dataset consists of daily epilimnion and hypolimnion temperatures. Simulations have been carried out using the semi-empirical OKPLM model. The authors also describe the model and its performance (including uncertainty and sensitivity analysis). The manuscript is clear and well written. Concise and nice to read, but probably too concise in some parts (see below). I have two main comments relative to sections 4 and 6, which are reported below:

- Section 4: Please explain what do you mean by "default parameter values" in the case of air2water and Flake. As for the first model, I do not think that default parameters are available. Also, it is unclear if the comparison discussed at lines 233-244 is against OKPLM run with default parameters (to be fair) or not. At line 229 the authors refer to 409 lakes, but in Fig. 1 and relative text to 401 lakes. Why there are 8 lakes more in the first case? Was the forcing used to run OKPLM, air2water and Flake the same? At line 231 I understand that air2water and Flake were run only with SAFRAN. Why the comparison with air2water and Flake was not done considering the period 1959-2020?

- Section 6: the authors should explain how they performed the sensitivity analysis and provide details about CSS. This should be anticipated in section 3.4. In their sensitivity analysis, the authors combine model parameters and forcing uncertainty (at_factor, sw_factor, MAAT). This looks a bit unusual: at_factor, sw_factor, MAAT are not model parameters. The analysis of Figure 4 is qualitative and should be improved: for which parameters are relationships between CSS and depth statistically significant? Are there statistically significant relationships also with other morphometric/geografical variables? The same considerations on the need to improve the analysis apply to Figure 2.

Minor comments:

L147 and L152: please, clarify what you mean by "an exponential smoothing function"

L155: please, provide information about the ALAMODE model.

L158-L163: please, provide the basic information to understand how the default "parameterization presented in Prats & Danis (2019)" has been derived. "The expression for epilimnion ..." and "the parameterization of hypolimnion ...": do the author mean "the values of the parameters"? Are expression and/or parametrization (i.e., equation) different compared to eq. 1-3? 

L188-195: I would restructure the paragraph anticipating lines 192-195 before lines 189-192. The authors say that S2M was used for simulating the water temperature in lakes situated at altitudes higher than 900 m, but in table 1 I see that some lakes at higher elevations have been run with SAFRAN.

Figures 1 and 2: I found the colour map difficult to appreciate, especially in the case of reservoirs (crosses)

L201: I am not sure what "initial assessment of the quality of OKPLM simulations" has been described in the previous section.

Section 3.4: the authors did not specify what objective function has been used (RMSE, NSE, MAE, other?). Is the range 0-1 for D, E and beta motivated by any physical/mathematical reasoning, or could it be wider?

Table 2: in the caption the authors mention a tilde, but they use a circumflex accent.

L226: I think the authors should refer to https://hess.copernicus.org/articles/17/3323/2013/ for air2water. 

Figure 3: please use capital letters for the parameters in the x-axis, as in the text. The font should be adjusted. mat should be modified into MAAT (also at line 265).