The paper analyses the main challenges foreseen by operational chains in integrating complex numerical frameworks. It also discusses on existing tools that facilitate orchestration of operational chain components. 

The paper serves as a guideline on the best practice that should be observed when setting up operational ocean forecasting systems. It also discusses main challenges that can be encountered in the operational chains. The paper is concise, well structured and avoids exploring too many details and thus the title, i.e. the use of word ‘Introduction’, reflects well the content of the paper. The paper also fits well within the scope of the State of the Planet journal. English is generally good although I do have several recommendations for improvement and several statements need revision and/or explanation.

In my opinion, the paper provides good, high level summary of the current state-of-the-art as regards the operational chains in ocean forecasting, makes reasonable recommendations and highlights main challenges that are usually encountered. Except for a few statements that are difficult to understand and/or with which I tend to disagree, and which, in my opinion, require clarification, I do not find anything controversial in the presented manuscript and I can recommend it for a publication subject to minor revisions, as per my detailed comments below.   

Ln. 29 remove ‘been’

Ln. 38 replace ‘facing’ with ‘to accommodate’

Ln. 39 replace ‘This chapter will give’ with ‘Chapter 2 provides’

Ln. 53 I think it is important here to stress that ‘different atmospheric forecasting fields’ must all come from the same atmospheric model. Also replace the word ‘fields’ with ‘variables’. It is not a good practice to mix atmospheric forcing variables from different models and I understand that the authors are not recommending it. At the moment it is not clear from the text and this needs to be emphasised.

Ln .54 replace ‘timeliness’ with ‘ time needed’

Ln 56 ‘and the time spent in having ….’ I do not understand what the authors are trying to say. Is it the same as ‘computational efficiency’ already stated in the same sentence, i.e. the time it takes to process this atmospheric data to make it ready for model forcing?

Ln 59 replace ‘In regional, rather than global’ with ‘Unlike in global models, in regional models …’

Ln 60 ‘that will be imposed …’ I do not understand this statement. Surely rivers will be implemented at the land-sea interface in the model domain, not at open boundaries. What does the phrase ‘of the limited domain’ mean here?

Ln 62 ‘is crucial’ . I disagree that it is crucial. It is desirable, not crucial.

Ln 65 replace ‘link’ with ‘linked’

Ln 65 ‘the observation assimilated using different class of data assimilation’ – I do not understand this statement. What is the ‘class of data assimilation’?

Ln 66 replace ‘cimmings’ with ‘Cummings’

Ln 72 replace ‘must’ with ‘can’. Depending on how raw data produced by the model looks and how it is served to end users, it ‘can’ be post-processed following the procedures described by the authors, but ‘does not have to’.

Ln 77 replace ‘among others’ with ‘such as’

Ln 78 replace ‘can’ with ‘may’

Ln 79 ‘including the use of a number of cores to produce the numerical solution that can be released in accordance with the correct forecast lead time’ I am not sure if I understand. Are the authors just saying that the number of cores used must be such that the forecast is produced on time? If so, then it can be phrased in much simpler language than it is

Ln 85 replace ‘lifecycle’ with ‘for’

Ln 91 remove ‘like’

Ln 92 add ‘for’ after ‘and’

Ln 98 change ‘others’ to ‘other’

Ln 103 move ‘ad-hoc’ to after ‘coded’

Ln 109 replace ‘lifecycle’ with ‘workflow’

Ln 180 rewrite the sentence to’ In order to properly reproduce spatio-temporal scale of the ocean processes, there is a need to:’