|First, I would like to thank the authors for considering most of our comments and suggestions and for their frank answers. I regret, however, that nothing was added in the introduction to highlight how the package fits among the many python packages dedicated to ocean analyses.|
Nevertheless, given the useful philosophy developed throughout the article and the valuable validation methods, the paper is worth publishing.
A few (~12) corrections and improvements are still requested and are listed below:
L31 Remove (see Section-??) and join the two sentences. "The Gridded class is used to read, represent and manipulate output from two NEMO model runs, and its use interactively with the Profile and Tidegauge classes allows a comparison between the model and observed data.
L34. Rewording: Information on the observation and modeled data used are given in Section 2.
L34-36 Please delete the paragraph, it is a conclusion, not an introduction. Also, "previous section" is non correct.
The sentence L34 "The package is open source and all of the code is freely available via Github (www.github.com)" could be replaced after L352.
The sentence "Using COAsT provides a level of transparency which aids knowledge sharing and discussion" could be moved to L366 after slight rewording.
L48 Even though the package aims to be extended to other ocean numerical models, it is not yet the case (see the documentation, the answer to reviewer 1, and the fact that analyses shown are based on NEMO only). Experiences always prove that despite the best prerequisites, extensions are not that straightforward. So please remove the reference to ROMS or at least add "in the future".
L67-L68 The sentence "It labels these dimensions with 1-dimensional time coordinates, 2-dimensional latitude and longitude coordinates and 3-dimensional depth coordinates." is unnecessary (repetition of L47). Please remove it. Moreover, the dimensions do not fully comply to the C-Arakawa grids.
L71. Could you please add a few words about the analysis classes? I am not sure whether the tidal and profile analyses are directly available from the COAsT package. It seems they are saved apart under https://github.com/JMMP-Group/NEMO_validation.
Move paragraph L65-L71 ("At the time... Table-2) to L62, just before "Analysis classes...
L82-L87 add no more information. Please consider just removing them ("All three of ... chunking and parallel computation")
L116 The three previous sentences are general and do not give any indication of what has been done in COAsT. After "module dependencies", consider adding a sentence stating that in COAsT, unit tests cover 67% of the package.
Section 3. The numbering is unbalanced. Why not numbering L180 as follows?:
3.1 Harmonic Analysis
3.1.2 Harmonic uncertainty estimation.. possible (L203)
L229-230 Please clarify.
- Was MHA applied at all locations where it was possible and basic HA everywhere else?
- Was harmonic uncertainty estimated at every location (even where MHA was possible) and then used to "mask" correct assessments?
L237 In the figureS
Figure 1. Again the numbering is confusing because Figure 1 is detailed after Figures 2 and 3.
Figure1/4 description. Please specify which method (MHA or/and just HA) has been applied. (see comment L229-230)