The authors present a unique LiDAR dataset of tropical forest sites on three continents including multiple platforms (terrestrial, UAV-borne and airborne laser scanning) and additional in-situ forest census data. The dataset is very valuable for the community for multiple reasons:

- Tropical forests are extremely rich in biodiversity. However, they are still understudied and there is less LiDAR and inventory data available than for forests in Europe and North America (which often have national forest inventories and national LiDAR datasets).

- Overlapping LiDAR data from multiple platforms allow for sensor comparison, derivation of reference data, calibration of sensor products (e.g., of ALS products via TLS data) and upscaling.

- The TLS datasets additionally include Quantitative Structure Models (QSMs), which enable the estimation of tree volume and aboveground biomass as well as reconstruction of virtual forest stands for radiative transfer simulations.

I thus recommend the publication of the dataset descriptor. Prior to publication, I suggest that the quality of the manuscript be improved and provide my comments below.

GENERAL COMMENTS

1) The dataset is very extensive with three study sites and three LiDAR platforms plus census data, making it difficult to review and also difficult to follow the manuscript and find relevant information. While I do support describing the dataset as a whole and including tha various sub-datasets, the authors are advised to improve the organization of the dataset (cf. 5. Data Access) and the description and presentation, e.g., by improving and adding figures and tables. This will make it easier for readers/users from different fields to find the relevant information and to access the data they are interested in.

2) The manuscript should state more clearly which analyses were undertaken and which can be addressed (by others) in the future – especially in Section 3 (Aligning and matching datasets). For this, it may help to modify the section title.

3) The authors should provide users with more information on the quality of the dataset (e.g., accuracies, error estimates or possible error sources, etc.).

SPECIFIC COMMENTS

1) It would be interesting to get more information about possible error sources as well as accuracy estimates, e.g., of alignment of scan positions and flight strips.

2) Do you have a way of judging the quality of the segmentation of the TLS data and the quality of the reconstructed QSMs? Also here, can you state limitations and possible sources of error? 

ForestScan Forest Biomass Reference Measurement Sites (FBRMS) 

3) Study site figures: It would be easier for the reader if all the study site figures would be coherent in style (font, colours, etc.) and content (overview map, map elements, legend, background map type to show elevation, etc.). Please find some further detailed comments below.

4) Consider addomg a world map that shows a marker for all the three sites on the three continents.

5) Fig. 1 would benefit from an overview map and a scale bar. It might also be better to mark the three ForestScan plots all by the same colour very prominently and label them by their plot ID directly in the figure.

6) Fig. 2 needs bigger labels, and a label for the colour scale would be useful. Also, it would be useful if the census and TLS plots as well as UAV areas were shown here.

7) Fig. 3: It is confusing that the legend mentions "4 ha" plots but the caption says "1 ha" plots, please clarify this. Please also add to the caption that there are always 4 plots in the corners of a "plot area".

Data

8) Tree census: I think a tabular overview or a timeline would make sense here so the user can quickly find the census times for each FBRMS.

9) Generally: If feasible, a timeline for all measurements (per study site) would be useful.

Terrestrial laser scanning (TLS)

10) Can you also add tables with scanner specifications (cf. Table 5 for UAV-LS)?

12) Pre-processing: Please mention and describe the downsampling of point clouds (cf. line 427), i.e., which algorithm and parameters were used. Was any filtering performed, e.g. in the RiSCAN Pro Import Wizard by reflectance or deviation, or outlier filtering? If so, please add this information.

12) Can you explain why the files are provided as .ply and not as .las/.laz?

13) Are additional point cloud attributes provided (e.g., Intensity/Reflectance, GPSTime, etc.)? When checking the dataset, I noticed that the .ply files have further fields, e.g., scan position index, reflectance, deviation and for the segmented point clouds the segmentation results and classification probabilities. This is very valuable information that should be included in the dataset descriptor.

14) Line 429: Is that classified point cloud also downsampled?

15) Line 443f.: "GNSS coordinates [...] for all scans …" → do you mean the coordinates of the scan positions, i.e., the scanner location? Please clarify.

16) Figure 11: So TLS almost consistently overestimates the DBH compared to the census? Can you state this explicitly in the text/caption?

17) In which coordinate system is the point cloud data provided?

UAV-borne laser scanning (UAV-LS)

18) Lines 499-508: Is this relevant for the data publication? Please either make the connection clearer (e.g., because flights have not been possible as planned due to restrictions) or leave this section out.

19) L. 522 / Table 6: maybe add an explanation like "a double grid plus an additional double grid at a 45° angle" to the term "criss-cross" and refer to Figure 12.

20) The plot IDs from the tables vs. in the text are confusing. Is FG6c2 the same as "P6". What does 200 and 100 mean in Table 6?

21) Table 6 and 7: Please make clearer which flights belong together and can be considered one acquisition (meaning that resulting point clouds are merged to one point cloud before further processing)

22) Please harmonize Tables 6 and 7, i.e., use the same "Date & Time" format (check ESSD author guidelines for preferred format), the same term for the velocity, and ideally the same specifications for the pattern (e.g., directions, interline distance, ..)

23) L 548ff.: Please improve the documentation of the LAStools processing steps and also explain what the mentioned commands do.

24) L 572ff: This paragraph is quite general and it is not clear what "must" or "can" be done and what was actually done for the specific published dataset. Did you use RiProcess?

25) L 582ff: Please make it clearer which steps CAN be performed and which WERE actually performed (e.g., ground filtering, individual tree detection). This is currently not clear from the text.

26) In which coordinate reference system is the point cloud data provided?

Airborne laser scanning (ALS)

27) L589ff.: This information could be summarised in a table, then it would be easier to find.

L617 – Section 3. Aligning and matching datasets

28) It is unclear if any of that was done. It should be stated more clearly what was done by you and what you recommend can be done with the dataset (by others) – Maybe also change the section header to something like "Recommendation ..."

29) L664-670: This sections seems unrelated to the alignment topic, please clarify (or leave out).

30) For further use of the data, can you refer to free software and tools to visualize and analyse the point cloud data (e.g., LAStools, CloudCompare, etc.)? This might be especially relevant for the QSM data, which is provided in a rather non-standard format (.mat files). Here, it would be interesting for the user, how they can open and further analyse the files (e.g., to export the tree models as .OBJ, etc.). It would also be helpful to guide the users how to transform the TLS data into georeferenced coordinate using the .DAT files and/or GNSS coordinates of the scan positions. 

31) Please add a "Conclusion" section as per the author guidelines (https://www.earth-system-science-data.net/submission.html#manuscriptcomposition) and check if a "code availability" section is needed.

32) Can you make it clearer that all entries in Table 8 are included in the dataset collection from the first row (if that is the case)?

TECHNICAL CORRECTIONS

General

1) Specification of coordinates: Please specify the coordinates always in the same format, i.e., either using minutes, seconds (like in lines 120-121) or using decimals (like in line 156).

2) Make sure to add spaces before units consistently (where applicable).

3) When referring to Figures and Tables in the text, consider omitting the word "below".

Tree census

4) L. 231 and 236: Is there the same DOI on purpose or is this an error?

5) L. 240: Do you mean "10 x 1 ha" here?

6) L. 244-245: The plot sizes are confusing here. Were there 9 x 4 ha plots and additionally 3 x 1 ha plots and 1 x 2 ha plot?

Terrestrial laser scanning (TLS)

7) Terminology: Decide for either "Terrestrial LiDAR Scanning" or "Terrestrial laser scanning"

8) Fig. 5:

8.1) Please call "Figure 5" and not "Figure 5a & 5b".

8.2) Labels for "upright" and "tilt" scan seem do not match the images.

8.3) Increase label font size.

8.4) Please, fix the labels in the caption, where 5b and 5c (which does not exist) are referred to.

8.5) The subplots a and b seem redundant (since only the labels differ). Can they be combined into one figure? 

9) Chapter "TLS data processing" (Line 333ff): Using Arabic numbers for the subsections might be confusing to the reader (cf. Section numbering), so please consider using letters (a, b, c) or Roman numbering (I, II, III) instead.

10) The TLS data acquisition section has a lot of repetition: Maybe the three sites can be summarized into one section (and the tables into one table)

11) Can you add acquisition dates to the tables?

12) Table 3: The caption seems incomplete ("Note: subplot 2 was").

13) Fig. 6: Add scale bar or colour bar legend.

14) Fig. 7: Add scale bar; add labels a) and b) and some descriptive captions. The colours are rather difficult to distinguish, can you adapt the style?

15) Figs. 8 and 9: It would be great if the point clouds and the derived QSMs used the same colour scheme for the tree instances, so clouds and QSMs can be better connected by the reader.

16) Table 4: Can you add the numbers from the caption to the table columns, so that it is clearer for the reader what you are referring to?

17) Figures 10, 11: Please increase the font size for the y-axis labels.

UAV-borne laser scanning (UAV-LS)

18) Table 5: Please add the datasheet sources as a reference in the caption. Please add the beam divergence to the table.

19) Figure 12: Please make the colour scale for the DSM in the legend continuous.

20) Table 7: Can you consistently use AGL and convert the amsl values to AGL in the table? (Alternatively explain to the reader why different height specifications are used here)?

21) Fig. 13: Labels a, b, c, d would be useful here and could then also be used correspondingly in the text.

Recommendation for data collection in FBRMS

22) L 704: 300 kHz is a specification of pulse repetition rate, not LiDAR power, please rephrase.

23) L 719: What does "harder to access" refer to, i.e., compared to what?

Data access

24) Tables 8a and 8b: Why not make it Table 8 and 9? Can you add columns for "Category" (i.e., Census, TLS, UAV-LS, ALS) and for "FBRMS no."? This would make it easier for users to find datasets. Are both URL and DOI needed or can the URL be omitted? Or even both, since the DOI is already included in the column "Citable as"? Please make the reference style in Table 8a and 8b the same.

25) You refer to "Section 4. data access" several times. Please correct this to "Sect. 5" (as per the author guidelines; please also check other section references).

References

26) Add access date to webpages included in the References.

Dear Authors,

Thank you for the opportunity to review your manuscript for possible publication in ESSD. I have provided a number of comments throughout, both as general observations and as specific notes referring to particular lines or figures. I hope you find these suggestions constructive.

General points

Title: consider this adjustment à ForestScan: a unique multiscale dataset of tropical forest structure integrating terrestrial, UAV, and airborne LiDAR with in-situ forest inventory data across three continents

Use of the word “census”: Consider changing this to “inventory” in every instance (also in the title), I find inventory is more commonly used in forest research and remote sensing fields. In any case you can use census as a keyword.

There are lots of authors, and I suspect lots of contributions – sections need to be unified in style and language. Language is generally good, but often sections could be more concise. Please consider

Acronyms: please check every acronym throughout the whole manuscript, there is a lot of repetition – acronym after the first instance only (e.g., DBH, TLS, ALS, AGB, EO, VOLS, ….).

Please pay attention to the technological limitations (e.g. accuracy of LiDAR – refer to Morhart et al. 2024 https://doi.org/10.1007/s10342-023-01651-z), this is touched upon in line 311. Also consider elements of quality assessment of data processing (e.g., wood/leaf class extraction as shown in Fig. 7). I would like to see a section devoted to discussing possible methodological/processing errors within the dataset. I would also like the methodological/equipment limitations to be discussed, this is touched on in line 271. I would argue that the tilted scan is nice to have but an extra step (and all the extra workload) that might not be essential (dependent on the size of the trees scanned). Did you scan outside the plot to better capture the trees on the plot boundaries?

Plot/sub-plot numeration becomes confusing between FBRMS and their individual subplots – consider making this more transparent/unified.

Consider adding a summary table in section 2 outlining the study sites, country, coordinates, area, climate, … this could considerably shorten the section which is a bit wordy. I would also like to know something about the stands that were scanned, tree density, mean DBH, top height, …..

Figures 1, 2 and 3 would be nicer if they had a unified style and similar content. For example, Figure 1 is really detailed and figure 2 is very vague. Some of the figures are missing basic mapping elements. It looks like they have been provided by separate people from the author consortium. Make sure the legends and captions are clear and correlate with the text body (e.g., lines 141 and 142 refer to treatment 1,2 and 3 – I guess this is T1, 2 and 3 in Fig 1? Also Line 230.

Mapping could also show the areas within each FBRMS that was scanned with each method. This is relevant, for example, to lines 589/590.

The methods sections (lines 281 – 332) could be condensed, call it data collection – the same or similar methods were applied for all TLS campaigns. One table for scanners used and their settings, one table for plot overviews. This part is very repetitive. E.g., Lines 318 and 328 (use of targets).

Line 766-775: Linking TLS trees to their census (inventory) counterparts. Plot marking is a valid point. Your examples are specific (e.g. Elephants). We too have tree tags that are destroyed by birds pecking at their shiny surface, we know the problem! We are also trialling QR codes on alu/foam board, plots can also be marked with a steel ground stake that can be found with a metal detector. The idea of anchor trees is good.

Specific points

Lines 44-46: define acronyms EO, AGB, cal/val, …

Lines 78-80: Better description needed – particularly on the need for better intercalibration.

Line 86: the acronym for Spacebourne LiDAR is not LiDAR. Also be careful in this section ESA Biomass is Radar not LiDAR – it has also already been launched (line 90), update needed.

Line 93: make à undertaken.

Lines 107-108: selection based on discussions is vague (people will always discuss), stick to the criteria.

Lines 112-114: Bullet points à note the acronym after the site name in brackets. E.g., Paracou Research station, French Guiana (FBRMS-01). Also see the general comment above about plot codes/names.

Line 122: define Cirad-UMR EcoFoG.

Line 125: thousand separators, additionally please check all.

Lines 135: remove “in the early 1980’s”.

Line 138: done à carried out.

Line 148: 9 à nine.

Lines 135-149: lots of (possibly) irrelevant detail that could be cut to make the section more focused (e.g., flux tower, fertilisation experiments). Please consider revising. Compare with the description of FBRMS 02 and 03 where there is much less.

Figure 2: a, b, c and d need defining in the caption (but consider first my general comment above).

Line 190: replace “laser-scanning” with “LiDAR” approaches

Lines 190-192: Sentence "species identity...." needs a rephrase. This and the next sentence can be made more precise. I think you mean the determination of tree species is critical since wood density x TLS derived volume = biomass.

Figure 4: remove – it is secondary information – just direct the reader to the right place.

Lines 201-202: Remove the sentence about ForestPlot.net – it is not needed.

Line 218: at 1.3m – this is DBH you have defined it before.

Lines 221-228: do you mean trees were recorded by their common names and then (as written in lines 226 – 228) trained botanists returned to identify species? If so try to combine these paragraphs.

Line 228: Explain APG IV

Line 233: Remove “referenced by a DOI”

Line 231 and 236: ask the journal how to display the hyperlinks, especially since the second is linked to a reference.

Line 240: Plot numbering/labelling confusion for the reader – link to map figure

Lines 238-241: much less detail shown here than FBRMS 01, why?

Lines 241 and 245: TLS was conducted not collected (TLS point cloud data was collected).

Line 242: “most” = vague.

Line 244: Same plot numbering problems as above.

Line 251: Gabonese and Malaysian FBRMS plots

Line 260: insert - chain sampling “protocols”

Line 261: QSMs needs describing

Figure 5a and b can be combined into one grid. Consider using axis labels to define scan position, the legend to the left must be much bigger to be legible.

Line 273: figure 5c?? missing.

Line 279: Working day is arbitrary – person hours give a better idea of workload.

Line 280: Maybe give the time needed for one scan.

Line 284: 16 quarter ha plots – that’s seems like a complicated way of expressing area.

Line 319: give detail on the RTK equipment

Tables 1, 2 and 3: the Lat. Long. coordinates are very approximate.

Lines 338-339: delete bottleneck sentence. It is true but not needed here.

Lines 351: Maybe specs are needed for HPC cluster and CPUs? Time reference (ca. 4 days) is ok here as the reader knows no break is needed.

Line 362: Precision of person hours vs. days (see comments above).

Line 366: “potentially containing more than 5.42 billion points” impressive, but too accurate and irrelevant to the reader, please delete.

Line 366: Insert à one small “exemplary” section of ….

Line 447, 552, 586, 599 and 616: delete “data access”.

Lines 455-459: please condense.

Figure 10: a nice figure but too much content, if only representative maybe just choose one plot row for e.g., 10 trees.

Line 476: an overview table might help the reader interpret the similarities and differences between flights (across all plots and flights) – Table 5 should be used and referred to earlier, and could be expanded for the other plots.

Line 477: VLOS stipulation is a repeat.

Line 480-481: irrelevant, I would not suggest anything other than adherence to the flight rules given.

Line 487-488: delete cherry-picker (above canopy platform is sufficient as a description).

Figure 12: legend entries need defining e.g., AOI, DSM

Table 6 and 7: is something up with the UTC date and time in the tables??

Table 6: define AGL (assumed above ground level).

Line 555: please rephrase and be specific that you are referring to FBRMS02.

Line 556: DELAIR DT26X drone platform

Line 561: “different” can you be specific?

Figure 13: Please label the panels and use fitting captions, I am not sure what all photos are showing me.

Line 609: ALS

Line 611: approximately = vague.

Line 620: WD, I don’t think this is the first instance?

Lines 664 & 668: rephrase – much as/far as possible.

Line 673: EO

Line 677: “This may sound obvious” too chatty.

Line 687: replace “types” with “sources”

Line 715: clearing à cleaning?

Line 716: what is (h)?