This dataset provides an integration of data from the World Ocean Database (subsurface ocean measurements) with data about the location of mesoscale eddies (as detected by satellite altimetry). This dataset is an exceptionally useful resource for amateur and experienced oceanographers alike, as it provides a well-documented manuscript description and high-quality data. I recommend publication of the manuscript and dataset in full in recognition of the novelty and usefulness of this product. 

The manuscript is thorough, well-written, and provides useful figures that demonstrate potential use cases of the dataset. It provides useful background about mesoscale eddies while focusing in the majority of the text on the dataset's creation and validation. 

Integrating Global Ocean Profiles Data and Altimetry-Derived Eddies

by Iury Simoes-Sousa, Cesar Rocha, Amit Tandon, and Andre Schmidt

--- Summary ---

This paper describes a merged eddy-profile database constructed by co-referencing World Ocean Database T and S profiles to a database of eddies identified and tracked using satellite altimetry. Since much of the sub-surface variability of the ocean is linked to mesoscale eddies, the collection of profiles along with eddy parameters should enable new directions of analysis for understanding eddy stirring and transport, dynamics and geography of eddy formation processes, and attribution of subsurface structure.

The approach has some similarity to others in the literature, including (a) SatGEM (mapping subsurface fields using regional binned profile averages by SSH; Meijers et al, JTech 2011), (b) SSH-Argo linear regression (constructing fine-grid relationships between SSH and subsurface T and S; Uriel Zajaczkovski Ph.D. dissertation, Scripps, 2017), and (c) eddy composite structure from satellite SST referenced to tracked eddies in SSH (Gaube, JPO 2015). This "eddy+profile dataset" approach is complementary to those, particularly in that facilitates further investigation rather than presenting conclusions.

--- General Comments ---

Overall, the manuscript is well constructed and well written, with compelling figures, and should need little revision. My comments are primarily suggestions about choices that would make the intended meaning clearer.

For example, I would suggest finding a different descriptor than "Eulerian" for the eddies tracked in SSH maps, since the most common use for Eulerian is to describe measurements fixed in space (e.g., moorings) rather than following the flow (e.g. drifters). Mapped SSH features are free to move around, even if they don't necessarily follow fluid trajectories. "SSH eddies" might be a better name. Or just "eddies" (following the nomenclature of Chelton et al, Prog Oceanogr. 2011) is fine too, as long as the methodology is clear. I realize the goal is to make a distinction from the Liu and Abernathey method which uses particle tracking in the time-varying SSH fields.

Plenty of useful detail is presented on both the prior datasets and the computational methods used to construct this eddy-profile database. Much of the WOD description is available elsewhere, but it doesn't hurt to include it here to remind potential users.

One point that could be made clearer in parts of the manuscript is the distinction between eddy observations on individual days and the identification of specific eddies with a lifetime over many days (e.g. the p.10 l.200 tallies of "cyclones" and "unique cyclones" seems to refer to single-day and full-lifetime numbers). A particular strength of this dataset appears to be that it has connected all profiles within the same eddy tracked over time (e.g. Fig 5), enabling study of an eddy's evolution over its lifetime.

However, since the algorithm connecting sequential observations into specific eddies is unique to the META3.2 DT eddy database (and presumably involves some choices that are made differently in other eddy-tracking algorithms), it would be useful to add some description of the algorithm here.

It's also a bit unclear exactly how much detail is available on each observation-eddy match. I'm guessing each profile location and the trajectory of each eddy center over time are all included (along with the eddy radius timeseries), so that composite eddy structures could be constructed. This is hinted at in Section 6 (Future directions), but a bit difficult to find stated explicitly.

--- Minor Comments ---

Title: suggest delete "Data"

p.10 l.195. It is mentioned that the dataset is separated into monthly files. How are specific eddies connected from one file to the next? Is each assigned a unique identification number that is carried through the entire dataset? Also, what is the relationship between the monthly "files" and 120 "chunks" that are processed separately. Is there a direct relationship such as 3 months per chunk (given 360 months in the 30-year dataset)?

Fig.5. Caption should specify that it (a) is the number of profiles over the full lifetime of each eddy and (b) excludes eddies with no profiles in them.

p.15 l.285. Some letters referring to Fig 7 panels are swapped.