The wet-bulb temperature (WBT) comprehensively characterizes the temperature and humidity of the thermal environment and is a relevant variable to describe the energy regulation of the human body. The manuscript proposed a global station-based daily maximum wet-bulb temperature data for 1981-2020, which is an important and exciting study for such a field. The GSDM-WBT dataset can effectively support the research on global or regional extreme heat events and humid heat waves. I think the study is well analyzed and did not have more comments. although I think the author should polish the expression of the manuscript. 

GSDM-WBT: Global station-based daily maximum wet-bulb temperature data for 1981–2020

Jianquan Dong, Stefan Brönnimann, Tao Hu, Yanxu Liu, and Jian Peng

ESSD

I really like what the authors have set out to do, to make a homogenised station-based dataset of T_w with complete temporal coverage.  I know that this is a complex task, and they should be commended for doing so.  They have validated their product against a reanalysis (which is as independent as one can get for these kinds of observations) and a similar station based product which uses alternative methods.

My major comment relates to the use of the NCEP DoE reanalysis, and how sensitive this data product is to the choice of this (older) reanalysis.  Most other comments suggest improvements for the figures, the readability or other minor clarifications.

Major Comment:

Line 95 - The NCEP DOE reanalysis is a relatively old product, and a number of studies have found issues with it. There are more recent reanalyses available (albeit being usually larger, these are more complex to use). How do you think your results depend on the choice of reanalysis?  

Minor Comments:

Line 12 - the phrase "and the response on human health" doesn't quite read correctly for me.  I see what you're trying to say, but that sentence doesn't flow well.

Line 16 - insert "stations" after "These"

Line 18 - it's not clear at this point what the offsetting mentioned does.  Is it possible to clarify a little (I appreciate it is the abstract, so space is limited).

Line 24 - "..resulting in the increase of the frequency...

line 29 - "response to"

line 31 - I think it would be important to indicate here that the "relatively low air temperature" refers to heat wave conditions, I initially read this as relatively low absolute air temperatures (which for me is 0-5C).

line 35 - I'm not sure this is the representation of the WBT on its own. However my understanding is that, combined with biophysical metrics, it can be used to determine whether sweating would still be effective.

Figure 1 - include in the caption a statement saying that the numbers in parentheses are the station counts left at each stage.  I'm not sure that "Correlationship" is a word (though I get what you're trying to say).  The arrow from the Complemented series back to Climatol is missing a head.

Line 82 - add suitable reference for the ISD (e.g. Smith et al, 2011)

Line 88 - how did you account for the shift to/from summer time when using the local time for the stations?

Line 121 - there is no section 3.1.1

Line 127 - restate that the 6-hourly intervals are in local time.

line 135 - "(of a total of..."

line 137 - "contain" (delete "s")

line 139 - insert a line break before "According"

line 156 - replace "for" with "so that there were"

line 160 - perhaps rephrase as "...depends on how many of the surrounding stations had missing data at this step"

line 165 - "needed"

Figure 2 - can you adjust the colourmap so that it diverges when centred around 0.  I appreciate there are few stations which have a bias of <0 but using a symmetrical scale will make that clearer than currently where blue-colours cover negative values and those 0-0.15C.

line 241 - move "in theory" to the front of that clause, to before "the corrected series"

line 262 - please check this sentence as there cannot be more than 30/31 missing days in a month.

line 272 - please add a suitable reference for the PHA algorithm

Figure 3 - similar to Figure 2, please can you adjust the colourmap so that it diverges when centred around zero.  Currently blue and pink colours are both negative, which could be confusing.

line 282 - please add a suitable reference for ERA5

line 283 - add parentheses around "2021" in the Yan et al reference.

Figure 7 - similar to Figure 2, please adjust the colourmap to center the divergence around 0.  Further guidance on this can be found at https://colorbrewer2.org. Currently it I found it difficult to identify the high and low bias regions.

line 314 - replace "index" with "indices"

line 351 - perhaps replace "congnizing" with "characterizing"

Figure S1 - add the number of valid and invalid stations to the caption.  The grey crosses are hard to see at first glance.

Wet bulb temperature is of great significance for the study of global or regional extreme heat events and humid heat, and this study realizes the calculation and homogenization of wet bulb temperatures at 1834 sites in a global long-term sequence, which is very meaningful work. Overall, this manuscript is clear and well-written and presents interesting research, but some concerns need to address before acceptance and publication. (Minor revision)

1. Figure S2 is difficult for readers to distinguish different zones. I suggest using different colors and symbols to distinguish different zones
2. As mentioned by the author, the air temperature, specific humidity, and surface pressure data of each site are reanalysis meteorological data of the nearest grid point directly extracted, but the spatial resolution of NCEP-DOE data is relatively low. When using the nearest neighbor algorithm to extract the temperature, humidity, and surface pressure data of multiple adjacent sites may be the same. Will this affect WBT calculation? If the bilinear interpolation algorithm is used for extraction, does the result of WBT change greatly?
3. Line 168: How is the initial daily maximum WBT calculated? The time resolution of NCEP-DOE reanalysis data is 6h instead of 1h. More detailed description is required here.
4. More detailed description of Figure 4 (a) is helpful for readers to compare the results before and after homogenization. In addition, “Before homogeization” should be “Before homogenization” in Figure 4 (a).
5. Line 243: What does higher SNHT value represent?
6. In Lines 264-265: The WBT is calculated site by site and day by day. The statistical results do show that there are many missing data of WBT in the HadISD-Humidity data, but the author believes that HadISD-Humidity has relatively low accuracy and higher uncertainties. From my understanding, the existing description is not enough to prove that HadISD-Humidity has relatively low accuracy and larger uncertainties.
7. As described by the author, the results of HadISD-Humidity and HiTiSEA are overestimated or underestimated, and the author uses HadISD-Humidity and HiTiSEA data to compare and analyze the results of GSDM-WBT, then the verification results do not represent the true accuracy of GSDM-WBT, but only the relative accuracy.