The paper presented a repository of thaw settlement test results sourced from the literature (generally sourced from Canada). The paper used the data in the literature to compare the effectiveness of existing empirical tools for estimating thaw strain and to identify the most fitting tools for various soil groups.

Although the tool presented in the paper will be useful for preliminary assessments, I'd like to point out that there is still a need for site specific data. If we don’t understand the underlying foundation conditions for the transportation infrastructure (i.e. example given in the paper), the estimates resulting from any of these methods may be completely off. This was demonstrated in Figure 12 where thaw strains are generally higher than the curve estimates. While the paper did not address it, there is also a possibility that water can infiltrate through thawing layers and exacerbate 'still frozen' layers, which in turn can affect the development of thaw settlements. The paper can be improved by removing some redundancy on the 'benefits,' which was repeated several times throughout the paper, and placing a strong statement either at the beginning or at the end of the paper.

The authors have used the context of climate change as a possible precursor for future issues with thaw settlements. It would be beneficial to the reader, and to the improvement of the paper, if the authors can highlight or provide an example how the data they have available in the PTSD can be used to 'predict' future settlements. For any of the data points available, assuming it was frozen at the time it was sampled at a certain depth, what would be the expected settlement today (if a temperature threshold is breached, and considering the recorded climate in the last 30 years)? 

Minor typographical errors:

Figure 1.a should be 2.a in Line 104

Line 179, Sec. 0?

This article aggregates and standardizes thaw settlement test data from a variety of sources, some of which previously unavailable, to allow for meaningful comparison. Additionally, the authors demonstrate the utility of this new dataset by benchmarking empirical estimates of thaw strain behaviour. 

Slightly more discussion of the quality control process would benefit a user of this data. More specifically, for data obtained using PlotDigitizer, is there an estimate on accuracy and precision? Similarly, in section 3, it would be valuable to elaborate on the quality control steps and the implications of the quality levels for someone who intends to use the data. For instance, under what circumstances (if any) could level 0 or 1 data be used? 

  

While it is mentioned in the paper, additional emphasis could be placed on the calculation of A0, a0, Cc*, e*th as explicit solutions to the problem of non-standardized tests. Making it clear to the reader early on that "These parameters form [the] basis for future comparisons of thaw settlement characteristics across various soil types" would serve to highlight one of the major contributions of this work. 

Finally, a number of specific technical changes will make the data easier to use, and improve the clarity of the manuscript:

Data and repository

-------------------

-In the git repository, consider mentioning which version of R was used for greater repeatability.

-borehole_location_and_datasource

    * do latitude and longitude values have an associated datum?  Is there one assumed?  Mention this in README metadata and/or in article

    * What is the estimated precision of lat/lon coordinates? Comment on this in S3.4 or S6. Some from 1976 are listed with six decimal places. Have these been re-surveyed?  

    * Values of ref column do not exactly match fields in the 'Citation' column of data_sources. This makes it difficult to automatically connect access details to boreholes to samples.  

-thaw_settlement_test_result

    * The 'unique_id' column might be better described as a 'sample_id' (also in sample_details and particle_size). Not only it is more descriptive, but 'unique' id in this context gives the impression that each row would have a different value which is not the case in this table or in particle_size

    *  I assume 'deformation' is 'relative to frozen sample?'  if so, add to README metadata description

-sample_details_with_derived_parameters

    * it appears at first glance that unique_id and sample_name form a one-to-one correspondance. If this is the case, and if you think it simplifies things, consider using sample_name as your unique ID (keeping the sample_name column). Otherwise, clarify whether unique_id and sample_name are indeed one-to-one in article

    * sample_volume_init is reportedly in mm3. However, values of ~800 mm3 with heights of ~100mm imply very narrow cross-sections. Is this correct?

    * sample_diameter_ave_or_typ :  units?

    * note: typo in "sewelling"

-data_sources_and_access_details

    * Are these reports accessible online anywhere other than the google drive?  Google drive links can be somewhat ephemeral.  One option could be to store the reports also in a more long-term repository (zenodo or other archive). this would depend on original authors. 

    * Consider omitting 'row' column or renaming to something more descriptive (source_id ? )

    * column names are not in same style as other csv files 

    * README file metadata incomplete for this csv file

    

Typos and Style suggestions

---------------------------

L11: "Infrastructures" in plural. Is this a typo?  Elsewhere in the paper it is always singular.

L68: Lewkowicz et al 2024 definition for thaw settlement: this glossary and definition is intended as a plain-language summary. Is there a more technical source?

L73: "obtained through this test" could be omitted for clarity

L80: in caption (and in other captions) references lack bold typeface

L82: "The thaw settlement [test] typically"

L90-95: I believe the notation (e - \sigma_v) is meant to represent the two axes of the plots described. However, it is easy to read this as subtraction when reading.  Is this notation common in engineering?  If not, consider explicitly introducing: e.g. "... presented as either thaw strain versus vertical (effective) stress (herafter referred to as a 𝜀 − 𝜎𝑣 curve) ... ", or using clearer notation. 

L136: missing period.

L179: "Sec. 0"

L194: "indicated by ODS in Figure 4" or omit parenthetical material

L287: consistency in units: negative exponent vs fraction for m3. Also period after kg

L325: can you be more specific about which of "the methods outlined in Sec. 2.2" you use?

L387: capitalize Appendix

L585: Consider adding a memorable named tag (e.g. 'essd-pts', 'to commit 5678ccd  (https://git-scm.com/book/en/v2/Git-Basics-Tagging)