General Comment.  

In this paper Zhou  al. present a large dataset compiling available moored observations of temperature, salinity and current velocities around Antarctica since the 1970s. The dataset is impressive, including close to 500 different datasets, covering a range of key ocean environments across all longitudes around Antarctica. Given the crucial role of Antarctic ocean and ice processes on climate, extensive re-use of the dataset by the research is warranted and may enable substantial advances in the field. I am sure the community will be very grateful to the authors for their great efforts to put the dataset together.  Therefore, I strongly endorse publication of the dataset and associated manuscript in ESSD. However, I have identified a series of potential issues and areas for improvement in the current version of the manuscript and dataset. It would be great if the authors could address my comments and suggestions, or at least respond to them, before I accept the article for final publication.  

Specific Comments 

My main concern about the dataset is that it is not very clear how the user would know how much trust they could place on each individual instrument, dataset. Where should the user refer to get information about if and how instruments (e.g. conductivity cells) were callibrated, how was the data treated between acquisition and publication, and about general data quality and flags (and/or whether instrument failure or drift are flagged). I also understand this is a complex task, and it may not be realistic to recover detailed data quality metadata from all deployments, but it would be great if the authors could talk about this a bit more in the manuscript, acknowledge this (important) limitation and say that users should refer to the original datasets, but also give an indication on whether the original files contain some more information. 

On a different note, it is great that the authors put together such a massive dataset. However, as I was reading the paper and having a look at the data, I have developed the feeling that the dataset can be a bit daunting, from the users’ perspective, due to its richness and diversity. I would suggest that the authors do some effort to further digest the dataset to give a clearer overview of it and make it more accessible. Some suggestions I could come up with are: 

• In the manuscript, present one or multiple figures and/or tables with  general statistics on the dataset: e.g. periods and seasons covered, record lengths, regions sampled, depths sampled, how many records with TS, velocities, both, etc. One good thing to report would be some statistics on the instrument vertical location with respect to local bottom depth / surface, since you mention there is a ”bias” toward deep measurements 

• In the datafiles. For each of the deployment files, you could add an overview in the metadata about the mooring location (not only latitude and longitude, but name of region), types of instruments and sampling depths and period. A link to the original dataset within files themselves could be useful. In general, richer deployment-specific metadata within each netcdf would be great. I am aware that modifying the whole dataset at this point could be a massive task, so please take this only as a suggestion, I leave to the authors’ judgement whether this is a sensible thing to do. 

• I wonder if the authors could share the code used to produce the figures for the preliminary analysis presented in the manuscript. That would give an example to users on how to bulk-access the data and generate some interesting insights from the ensemble of observations. 

In line comments 

Line 98. Maybe provide an early indication of the size of the dataset: e.g. “This compilation includes 521 mooring time series [...]” 

Line 109. Could you include some general description of the SOOS mooring dataset and how many more records you are including in your database. 

Line 114. There is a typo here “,.” 

Figures 3-5. I felt the scatterplots may look better on log-colour scale, to highlight overall patterns rather than some particularly high values in some locations? Not sure... Also, I suggest annotating the maps with key locations mentioned in the text, e.g. on line 227 and others. 

Lines 201-214, Lines 222-232. I think these paragraphs would benefit from more support from citations. 

Line 212. “... mixing ...” Turbulence may be a more appropriate term here 

Figure 6. In the caption, can you add a bit more information about the different features presented in the TS plots here?