Peng et al., presented an important data set for understanding permafrost degradation and the corresponding consequences in Qilian Mountains. The data set is unique and useful, but the writing of this manuscript needs to be improved for clarity. Please see my comments below,

Major comments:

The data set (shapefile) only has two attributes (O_name and Areas). What does O_name mean? Could you add more attributes: the person who identified the features, The type (RTS? Active-layer detachment, and thermos-erosion gully), the acquisition dates of satellite imagery for identifying the features, the source of satellite/UAV imagery, the initiation time (year) of the features, and so on?

The data set contains 2717 polygons, but the manuscript said there are 1064 HT features, how does each polygon correspond to each HT feature?

When you analyzed the spatial distribution and characteristics of these features, could you group the features based on their types: retrogressive thaw slumps, active layer detachment, thermo-erosion gullies, and then conduct the analysis separately? Different types of hillslope thermokarst are quite different in shape, geomorphology, and triggering factors. For example, gullies have narrow and elongated shapes, but you concluded that all features are not elongated.

Minor comments:

L32: shallow active layer depth? What’s the depth of the active layer?

L49: “instoring” to “in storing”

L57-59: Olefeldt et al., 2016 focused on the Arctic and didn’t have any data for Tibet. Do you have a better citation for this sentence? How about other works based on field investigation and remote sensing? I believe there are several published papers related to this statement.

L60-61: permafrost lake? Change to “thermokarst lakes”?

L67: what are “Google Earth satellite images”? By the way, Lewkowicz & way, 2019, used Google Earth “Engine”.

L69: “circumpolar Alaska“  What does this mean?

L83-84: Qilian Mountains is only a sub-region of QTP, how it affect the ecology of the QTP? This sentence is confusing.

L89: “in the region”, which region? you were talking about Qilian Mountain but cited a paper for Alaska.

L90-91: what are “terrain features”? By the way, when you identify hillslope thermokarst, how did you distinguish normal landslides (not caused by permafrost degradation) and those features caused by permafrost degradation?

L99: do you have specific values of the mean or max of the annual precipitation? Otherwise, how do you know the variation of the precipitation was high?

L104: “security” of what?

Figure 1: What’s the white region in (a)? lake? unfrozen ground? Others? Could you mark the locations of features in (b) to (h) in (a)?  Could you also name the features (e.g. gully or RTSs) in (b) to (h)? In (a), you only show the extent of permafrost. Could you show the extent of continuous, discontinuous permafrost?

L117: what is Omap? Could you provide a link or citation to it? I think most people don’t know this, unlike Google Earth.

L120: tabulate? Change to calculate?

L122-124: do you have a citation for this formula?  How did you define the region of interest? What’s the size of the region of interest? What’s the physical meaning of the TPI?

L138-139: here “2 m” mean 2 m above ground? Not their spatial region is 2 m? right?

Figure 2: what’s the implication of the latitudinal and longitudinal distribution?

L162: what do you mean by “thermokarst landslides”?  RTS? Active-layer detachment slides?  Or Others? What is the difference between “thermokarst landslides” and those landslides not in permafrost regions?

L165-168: I don’t understand this sentence.

L174-176: a situation where an HT feature already existed for a long time then captured by satellite imagery. Is this still ok?

L190: “al”?

L192: “Shiyanghe basin” and “Datonhe-Huangshui basin” are not in Figure 4.

Table 1: how do you know imagery resolution in Google Earth Pro is 0.6 m? As I know, the resolution could vary in different locations and time, and the data source could also include others such as WorldView 1, 2,3, and 4.

What is “Esri Wayback Imagery”? The source could also be some satellites.

L204-208: what’s the value range of LSI?

L212-215: I don’t understand this sentence.

L221-228: what are the purposes of the autocorrelation?  The correlation between what and what? This part is confusing.

L234: How do you calculate the average here? Spatial? Or temporal? Why need the average?

L259-260: this sentence mentioned HT, but Luo et al., 2022 only focused on retrogressive thaw slumps.

L268-269: “most HT is simple in shape and compact in morphology, rather than elongated”. If this is a statement for retrogressive thaw sumps, then it’s ok. But HT also includes thermo-erosion gullies, which are usually narrow and long.

L270: and you also cited Yang et al., 2023, which only focused on retrogressive thaw slumps.

L276: “bordering the Heihe basin”. I don’t understand.

L336: why did earthquakes cause permafrost warming?

L324-337: the analysis of the seismic data is quite weak. You know the timing of earthquakes, but how do you know the initiation time (iin hours or minutes) of these HT features?  The only information is the spatial distribution of HT features and earthquake epicenters. As shown in Figure 7,  there are clusters of earthquake epicenters in the lower left, but no HT feature there. How do you explain this?

Figure 6 is simple but needs to be clarified.  How do you define extreme weather events based on the ERA5 data? What kind of extreme events? The horizontal line represents the mean of what?

L359: where are these peaks?

Xiaoqing Peng et al present the first dataset about the hillslope thermokarst in Qilian Mountains. The data presented are useful for the cryospheric hazards study. It could provide as the important samples for the deep learning in the permafrost regions. Meanwhile, it is also useful for the sustainable development for the local regions. For these reasons, this database should be of interest to many aspects, e.g. permafrost degradation study, environmental impact assessment, deep learning etc. However the manuscript does require a number of revisions before it can be accepted for publication.

Data:

1. The file name of the data seems not normalized, e.g. ‘shp’. I suggest the Polygons revised as HTs.
2. The meaning of HTs_V1 seems not clear, which could be as HTs_Shapefile.
3. The data “read me” (file description) file needs more detail and clear for readers, e.g. ‘"O_Name" field indicates the time of the satellite imagery for this morphological element,’. These should be more detail for the name of satellite.

Specific Comments

1. As we know that, the thermokarst are mostly used or defined in the permafrost regions. However the hillslope thermokast is unusual. So, please explain it in detail for hillslope thermokarst.
2. The sentence in line 103: "Permafrost instability in the Qilian Mountains has gradually increased, resulting in HT formation including RTSs, thermokarst lakes, and thermal erosion gullies" is incorrect. As thermokarst lakes are not included in the hillslope thermokarst hazard.
3. I suggest changing "freeze/thaw-induced hazards" to "hillslope thermokarsts" in line 84.
4. Line 1 invertony should be as inventory.
5. Please note that there are some English grammar and presentation problems in the article, as well as formulas in lines 122, 204-205 that do not appear properly in the manuscript.
6. Lines 600-602 do not have DOI added to the citations of the reference, please uniform the citation in manuscripts.
7. I think you have some problems in section 4.1, such as the lack of precision and clarity of "HT", "thermokarst landslides" and "landslides" in lines 161-163.
8. The Jilin-1 satellite image you mention in "7 Data Availability" does not seem to be used in "4 Methods" or other chapters, and the authors should make a clear statement as to whether or not this data is used, and how it is used.

12. Please carefully check the format of article, for example line 358 “HTwas”, line 414 “mid- and upstream”.

1. In lines 261-262 of the manuscript, the authors state that 90% of the RTSs tend to be distributed in the altitude range of 3200-4000 m. Is this statement accurate? In context, it would be more standardized to use "90% of HTs" instead of "90% of RTSs".
2. In the study, there are several river basins, but it does not appear in the figures, please added it.
3. About the potential influence factors, it is simple for the impacting factor earthquake, which could be more detail for explaining. For example, it can be seen that there are some fissures after earthquake, which could be the important factors controlling the HTs development.
4. Line 120, the word tabulate is not suitable.
5. Line 190, what is ‘al’?
6. In the discussion, the climate factors does not show detail in the methods.