the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 May 2026
High-resolution Digital Outcrop Dataset of a Fossil Hyperextended Rifted Margin, Swiss Alps
Abstract. Structure-from-motion (SfM) photogrammetry provides new possibilities for the interpretation of complex geological objects and settings through the digitalization of outcrops in the form of Digital Outcrop Models (DOMs). This study focuses on the acquisition, processing, and georeferencing of 12 DOMs targeting the Err and Bernina low-angle normal faults (LANFs) in the Central Alps. This area exceptionally preserves remnants of Jurassic rifting. Extensive Unmanned Aerial Vehicle (UAV) -based field campaigns (2022–2025) over 43 km², covering 1710 m elevation difference, combined with differential GNSS and regionally-available surface data, produced 12 high-resolution DOMs and associated products (point clouds, textured meshes (DOMs), tiled models, orthomosaics, and DEMs) with centimeter to decimeter resolution. A total of 15 control points per DOM were used to georeference and quality assure the digital data assets, 5 of which function as reference check points (CPs). Within the twelve DOMs, the total ground control points (GCPs) root mean square error (RMSE) ranges from 0.02 to 1.47 m and the RMSE of the CPs ranges from 0.49 to 3.11 m. Individual DOMs within the Fossil Alpine Tethys rifted margin DOM (FATDOM) Dataset reveal detailed internal fault structures, lithological variations, fracture networks, and tectono-sedimentary relationships, offering new insights into the architecture and kinematic evolution of LANFs that also extend to the seismic scale. Comparison of our DOMs with seismic data in present-day systems can be used to bridge the scale gap between local structural observations and regional interpretations. Beyond tectonic implications, the high resolution of our resulting DOMs enables a wide range of geoscientific applications including geomorphological studies focused on the monitoring of deglaciation. The data described in this paper are available on Zenodo under https://doi.org/10.5281/zenodo.18940068 (Morzelle et al., 2026).
- Preprint
(5335 KB) - Metadata XML
-
Supplement
(7 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on essd-2026-194', Anonymous Referee #1, 27 Jul 2026
-
AC1: 'Reply on RC1', Leïla Alili Noah, 17 Sep 2026
Dear Reviewer,
Many thanks for your constructive comments, which we have addressed in the revised manuscript and which have helped improve the overall quality of the article. Please find the responses to each comment, highlighted in red, in the attached PDF.
Best regards.
Leïla Alili Noah and co-authors.
-
AC1: 'Reply on RC1', Leïla Alili Noah, 17 Sep 2026
-
RC2: 'Comment on essd-2026-194', Samuel Thiele, 29 Jul 2026
The paper “High-resolution Digital Outcrop Dataset of a Fossil Hyperextended Rifted Margin, Swiss Alps” presents an impressive (in scale and extent) and interesting set of digital outcrop models covering well exposed low angle normal faults in the Swiss Alps. The combination of interesting geology and alpine geomorphology (including several landslides and glaciers) will make the dataset of interest to a diverse audience in both research and teaching.
All of the data (including, importantly, the raw drone photographs) are available and well documented in FAIR data repositories. I would thus suggest that the manuscript could be accepted in ESSD following the minor revisions outlined below.
General comments:
1. The start of the paper could emphasize more the diverse applications of this dataset (structural geology, sedimentology?, geomorphology, glaciology, teaching?) to ensure it reaches a diverse audience.2. The description of low angle normal faults could be improved, and ideally written for a somewhat more general audience (matching the previous comment) — especially given their relative novelty and impressive scale.
3. If possible, please also include validation / field data (e.g., field notes, compass measurements of structure orientations) in the open dataset, as these are invaluable if used for further research.
4. Please revise the manuscript to tighten the language/writing and ensure clear and engaging narrative flow.
Specific comments:
Line 80: Please forgive my ignorance, but how did these low-angle structures avoid reactivation during alpine compression and associated thrusting? I imagine the explanation is worth at least a sentence or two.
Line 95: Why manual focus? For most UAV cameras auto- or fixed (infinite) focus should be sufficient?
Line 99: Is 0 degrees vertical (NADIR) or horizontal?
Table 1: Please add a scale corresponding to the approximate GSD (calculated from camera to outcrop distance) to the bottom of the “Distance camera/outcrop (normalized)” column.
Line 125: Delete this paragraph ;-)
Line 136: It would be worth commenting here on the georeferencing accuracy of the SwissALT3D products?
Line 230: Clarify that some (many?) of these deviations could be real topographic change?
Line 295: Double check this - I would imagine the errors you report are nearly 1 order of magnitude higher than a typical RTK direct-georeferencing survey (30cm to 1 m vs ~1-30 cm).
Line 332: Consider mentioning here the benefits of converging imaging geometries (e.g., flying around a target imaging obliquely) vs planar or diverging views (errors tend to cancel during the bundle adjustment, improving the robustness of the solutions; see some of the papers from Mike James on the role of imaging geometry).
Line 365: This sentence could be reworded for clarity: how does limited variability in color or texture reduce risk of incorrect interpretations? I don’t follow the logic.
Line 425: This whole paragraph would fit better in the results section rather than the discussion.
Line 485: Are there other photogrammetry based studies / datasets covering these glaciers? Worth noting / briefly reviewing these if so.
Line 513: Also mention CloudCompare?
Further minor suggestions are included in the annotated PDF.
-
AC2: 'Reply on RC2', Leïla Alili Noah, 17 Sep 2026
Dear Sam Thiele,
Many thanks for this encouragement and constructive comments, which have helped improve the quality of the manuscript. Please find the responses to each comment, highlighted in red, in the attached PDF.
Best regards.
Leïla Alili Noah and co-authors.
-
AC2: 'Reply on RC2', Leïla Alili Noah, 17 Sep 2026
-
RC3: 'Comment on essd-2026-194', Brian Burham, 22 Aug 2026
The manuscript “High-resolution Digital Outcrop Dataset of a Fossil Hyperextended Rifted Margin, Swiss Alps” presents a comprehensive and spatially extensive dataset of digital outcrop models. The breadth of potential applications of the data is great to see presented (structural geology, geomorphology) and will be useful to many researchers in related fields. The raw imagery and ancillary documentation is well-organised, thorough, and FAIR-aligned – a critical part of science and something that digital outcrop data is well positioned to provided. So it is great to see significant strides being taken in this. The geometry quality assessment is well handled and the accompanying figures are supportive.
Though the data is good, the presentation of it could use clarification to make it easier to validate and useful for practitioners along with the language use to make it more direct and ingestible.
General Comments
1. Reduce first-person voice with more direct third person phrasing.
2. Consider using more consistent terminology. For example, GSD defined once could be used throughout the manuscript instead of phrases like “ground resolution” or “cm/pix”.
3. I think the manuscript would benefit from referring to models by name instead of number. Using model names would be more useful to readers who are not familiar with the convention and may want to use associated data that they may have.
4. Brief clarifications of some technical terms would help readers and practitioners to grasp what information is presented and then use it more readily. E.g. “confidence” or “internal error analysis”.
5. I suggest the authors reconsider the deglaciation monitoring section. I feel it may sit outside of the scope of the main purpose of the paper, and it is a substantial area of research in its own right. Though I think the DOMs and associated DEM data is great for this use case, it may be better suited in a separate manuscript.Detailed Comments
L1: “Fossil” - Is Fossil the right word here? I think simply removing 'Fossil' would make the title more impactful.
L22-24: “Beyond tectonic implications, the high resolution … the monitoring of deglaciation.” - This part of the abstract is not needed and is outwith the scope of this manuscript. See related comments at associated section.
L33: “2024;” - Replace with “2022 ?”.
L40: “Their characterization is” - Insert “accurate”.
L44: “fossil” - Delete.
L48: “exceeding 2 500 m.” - Insert “,”.
L49: “of these sites. As a result,” - Burnham et al., (2022) discuss this. Consider adding a citation here.
L49: “the” - Delete.
L54: “fossil” - Delete.
L55: “Fossil” - Perhaps "Ancient" is a better word?
L57: “findable, accessible, interoperable, and reusable” - This should be placed before the FAIR acronym so that it is properly defined.
L62: “unique fossil example of” - Delete.
L85: “acquisition” - I understand what you mean, but to be more precise, which is important in a paper like this, to state that the UAVs enabled imagery acquisition which facilitated photogrammetric model construction and generation of DEMs...
L90: “We conducted four UAV” - The authors should consider not using 1st or 2nd person voice and write a bit more directly. perhaps something like "Four UAV flights were conducted..."
L93: “as” - Delete.
L94: “various” - Delete.
L95-96: “while manually re-focusing was done in between.” - Replace with “Manual camera lens focusing was completed between each image”.
L97: “while avoiding” - Replace with “to minimize”.
L119: “Textures” - Describing the texture resolution and resultant number of textures for each model would be useful to readers and users.
L122-127: “Figure 1 uses the style Caption. The abbreviation … the colour schemes accordingly.” - Please remove the author guidelines text from the manuscript.
L138: “the model 11 integrated from Betlem et al., (2023a)” - The authors should consider describing this model (and all others throughout the manuscript), at least by name, would be more beneficial to readers.
L143: “the total” - Delete.
L159-160: “5 to 15.4 cm/pixel” - The sentence structure makes it sounds like you determined the DOM pixel resolution the same as orthomosaic resolution - but I do not see that in Table 4. I can understand the orthomosaic and DEMs having a resolution as you describe, but DOM resolution, as the authors know, is difficult to quantify because they can vary depending on triangle number and coverage. Correct the sentence structure if this is a typo, or if the authors did determine DOM pixel size, a description of how this was achieved would be useful.
L169: “Figure 3:” - I really like the colourmap choice on this plot. Linear and perceptually uniform colourmaps are a more accurate representation of the data and it is colour vision deficiency safe!
L173: “elevation values ranging from 2 680 m to 3 250 m,” - The authors should consider adding the contour interval value to clarify what the difference between successive contour lines are.
L175: “Processing of the datasets” - Swapping the noun and verb here would make it more succinct and remove passive voice. Perhaps something like: Dataset Processing
L178: “Models 1, 2, 3, 10, and 12” - I think providing the names of the models instead of their number would be beneficial for users who will have that model in mind and look for the associated data. This should be completed throughout the manuscript.
L193-194: “owing to how the calculation is made per horizontal square meter.” - I'm not sure I understand what is being stated here. A bit more description of what you mean is needed here. Presumably, the UAV pilots focused on the geological areas of interest (outcrops) and less on the periphery. Therefore, there are more images which allows the SfM and MVS algorithms to identify more common points. Surely the pts/m2 is
L196: “average confidence” - The authors should consider defining confidence a bit more thoroughly here - confidence could mean a few things in this context.
L198: “Both” - Delete.
L198: “highest density and” - Delete.
L201-202: “as also evident from the reduced point density and camera overlaps” - I recommend this be discussed in the relevant section about point density (C in the charts).
L206: “cm/pix” - The authors should consider using the consistent naming conventions or abbreviations. Either is fine, as long as it is consistent.
L206: “low” - Consider rephrasing to use 'small' instead of low'. The word 'low' could be interpreted as elevation in this context with regards to DEM and orthomosaic.
L213: “(internal)” - Could the authors clarify what an internal error analysis is? This would help others reproduce the error analysis.
L230-231: “covered from a greater distance (Figure A9), as does model 4 in the southwestern part,” - The authors should consider rephrasing to clarify that 'greater distance' refers to UAV distance from outcrop. At the moment, 'covered from greater distance' does not make sense in this context.
L260: “quantifiable data form tool suited” - The authors should consider rephrasing the sentence, because at the moment it is unclear what the tool is and what it is suited for?
L262: “fossil” - Delete.
L273-274: “avoiding control points located in areas subject to seasonal or short-term surface changes,” - Excellent! This is a key point users should be aware of.
L277-278: “with ice dynamics and retreat were also avoided, as … georeferencing accuracies.” - The authors should consider rephrasing this for clarity. Perhaps something like "Additionally, glacier-covered areas present in some models were avoided because the elevation and geometric temporal variability, as a consequence of ice dynamics (e..g, glacier retreat) can affect the overall geospatial accuracy."
L280: “display reduced reliability because” - Insert “geospatial”.
L282: “likely.” - Replace with “common”.
L282: “xyz errors” - Is this 'xyz error' an average of the three vectors? A simple clarification sentence would be a useful addition for references to this type of error in the manuscript.
L298: “our” - The authors should consider refraining from using 1st or 2nd person voice. I would suggest just using "the" here
L300: “dataset we acquired.” - The authors should consider, as above, from using 1st or 2nd voice. Instead, using something like this that is a bit more direct and doens't use passive voice "acquired dataset"
L308: “ground resolution” - In aerial surveyors and remote sensing scientists typically use Ground Sampling Distance (GSD) when referring to accuracies like this. The authors should consider using this terminology for consistency.
L308: “resolution” - Replace with “GSD”.
L323: “our” - Replace with “the acquired”.
L324: “our” - The authors should consider using terminology here, to reduce the use of 1st and 2nd person, to something like "FATDOM"
L336: “our” - I would encourage the authors to simply reference the DOM dataset itself in this context because it is different from the SwissALTI3D data, therefore does not require the discriminator. This also removes the 1st or 2nd person voice.
L339: “the” - Delete.
L340: “(around 0)” - The authors should consider stating what the observed error levels are.
L346: “variations and limitations of the datasets” - I would suggest the authors choose subsection titles with that are more direct and do not use the passive voice. Perhaps something like "Dataset quality and limitations"
L348: “ground resolution” - Replace with “GSD”.
L357-359: “In addition, the observed distortions (e.g. image … (e.g., see model 8; Figure A8).” - The authors should consider moving information from this sentence to the sentence what discussed the Blurred areas, one sentence above.
L361: “Textureless domains” - The authors should consider adding information to clarify what a textureless domain is. Presumably these are areas where insufficient texture information is available. Stating something like this is sufficient.
L365: “is small,” - Insert “.”.
L366: “for” - Replace with “F”.
L369: “spatial resolution” - I assume you mean GSD here? I would suggest that the authors simply state that since it would have already been defined previously.
L385: “cliff scale” - Could a cliff not also be an outcrop? I'd suggest that the authors do not need the latter half of the sentence and simply state that the literature presents numbers DOMs were acquired in the Alps.
L389-391: “This resolution is one of several key criteria for … and stratigraphic features.” - This is true and a key point - are there references that the authors have as examples or back up the statement?
L400: “FAIR principles” - This indeed crucial. The authors should consider a new online platform that was built around FAIR principles specifically for DOM sharing. It has the ability to link publications, published datasets, and even compose papers/internal documents/field guides that are linked directly to outcrops published. https://outcrop.io/
L400: “(findable, accessible, interoperable, and reusable)” - Delete.
L407: “area” - Replace with “areal”.
L418: “Usability of the dataset” - The authors should consider rephrasing the Section title to be a bit more direct. perhaps something like: "Dataset Utility" or similar.
L422-423: “Finally, the FATDOM data can also be used for other … and mapping of Alpine structures.” - The authors should consider add a few references of studies that use DOMs for glacial monitoring.
L425: “Our” - Delete.
L443-444: “Thanks to the DOMs, the internal structures of … be visualized and investigated.” - The authors should consider rephrasing this to state that DOMs facilitate visualisation and investigation instead of thanking them. Perhaps something like: "DOMs facilitated visualisation and investigation of allochthonous blocks internal structures that are composed of pre-rift sediments and basement rocks."
L444: “by” - Replace with “in”.
L444: “the” - Delete.
L445: “on the virtual outcrop.” - This section is not needed, it is already implied by the first part of the sentence when you use VRGS. The authors should also be mindful of consistence when using digital outcrop model or virtual outcrop model.
L449: “Our” - Replace with “The”.
L449: “Our DOMs provide a” - Insert “also”.
L455: “Figure 5:” - Great figure! I like how the authors have visualised and illustrated the different data collected on the DOM.
L461: “The” - Replace with “DOM”.
L462: “although” - Replace with “however,”.
L463: “fossil analogue of rifted margin” - Replace with “rifted margin analogue”.
L464: “new information” - Replace with “insights”.
L464: “leading” - Replace with “that lead”.
L465: “of such comparison” - Delete.
L465: “It illustrates” - Delete.
L474: “imaged” - Replace with “identified”.
L479: “Use in geomorphology: monitoring deglaciation” - I'm not sure this section belongs in this manuscript. Glacial monitoring with DEMs (from any source) is a large area of research that is outwith the scope of this manuscript. This type of analysis, however, could merit a separate manuscript, if the authors wish to conduct further analyses. I would therefore suggest the authors consider removing this section and associated figure.
L513: “Buckley et al., 2022.” - Not associated with VRGS
L616-618: “Burnham, B., Bond, C., Flaig, P., Van Der Kolk, D., … 2024.” - Delete.
Figure 3. The authors should consider adding a label to the top right of each chart to quickly highlight what the chart is showing. This would allow users to quickly identify what the chart they are looking at, or looking for, is showing. I suggest this is added to similar and relevant supplementary figures. i.e., 'overlap', 'orthomosaic', 'point density', 'confience' etc..-
AC3: 'Reply on RC3', Leïla Alili Noah, 17 Sep 2026
Dear Brian Burnham,
Many thanks for this encouragement and constructive comments, which have helped improve the quality of the manuscript. Please find the responses to each comment, highlighted in red, in the attached PDF.
Best regards.
Leïla Alili Noah and co-authors.
-
AC3: 'Reply on RC3', Leïla Alili Noah, 17 Sep 2026
-
AC4: 'Reply on the Editor comments', Leïla Alili Noah, 17 Sep 2026
Dear Robert Jackisch,
Many thanks for the constructive comments and the time spent. The comments have been addressed in the revised manuscript and helped to improve the overall quality. Please find the responses to each comment, highlighted in red, in the attached PDF.
Best regards.
Leïla Alili Noah and co-authors.
Data sets
FATDOM dataset Leïla Morzelle, Peter Betlem, Geoffroy Mohn, and Julie Tugend https://doi.org/10.5281/zenodo.18940068
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 368 | 99 | 41 | 508 | 28 | 35 | 41 |
- HTML: 368
- PDF: 99
- XML: 41
- Total: 508
- Supplement: 28
- BibTeX: 35
- EndNote: 41
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General Comments:
------------------
The submitted work presents a dataset containing several digital outcrop models generated by UAV photogrammetry.
The main objectives of the dataset are to preserve outcrops in a rapidly changing climatic environment and to improve the understanding of low-angle normal fault formation processes..
Data acquisition and processing are described in considerable detail. The resulting DEMs and orthomosaics are evaluated with respect to their geometric quality using control points and official elevation models.
In addition, the authors present several potential applications of the dataset. The dataset is publicly accessible via Zenodo, supporting its findability and accessibility.
Significant & Data Quality: Good
In my view, the dataset is significant with regard to the stated objectives, and the overall data quality appears to be good. However, I can assess the broader geological significance only to a limited extent, as my background is mainly in photogrammetry.
Presentation Quality: Fair
My main concern relates to the presentation quality. Parts of the workflow are written in the style of a technical report or software manual. Some steps remain unclear, and the motivation for applying particular processing steps is not always sufficiently explained. The Discussion contains several statements that remain unclear or open to debate. It would benefit from a stronger focus on the main findings presented in the Results and from a clearer discussion of their implications. A reorganization of some subsections may also improve the overall structure and readability.
Specific comments:
------------------
1. The level of methodological detail appears somewhat unbalanced. For example, the manuscript reports detailed information on intermediate products, such as the number of points removed during sparse point-cloud filtering, while important information on the camera sensors is missing.
Please provide the relevant camera specifications, including at least the sensor dimensions and image size in pixels.
2. The description of the photogrammetric workflow is partly too closely aligned with the sequence of processing steps in Agisoft Metashape, while the methodological purpose of these steps is not always sufficiently explained.
For example, around lines 130 ff., the procedure for establishing the GCPs is described, followed only briefly by the phrase “prior to tie-point filtering.” At this point, several important details remain unclear.
Please clarify what is meant by “Post-Processing Kinematic (PPK) data.” Does this refer to PPK-corrected GNSS camera positions, and how were the underlying GNSS observations acquired and processed?
Furthermore, the origin and use of the natural markers should be described more clearly. How were these markers identified, how were their coordinates determined, were they manually selected?
Another example is the filtering of tie points. Here, the text should (shortly) explain the purpose of the applied filtering steps, rather than only naming them.
3. I suggest moving some aspects of the discussion into the introduction and motivation. In particular, the manuscript should more clearly distinguish the present work from previous studies, identify the resulting research gaps, and briefly summarize the raw data and derived products provided in the dataset.
4. Section 5.3, “Usability of the dataset,” might be better placed outside the Discussion. The presented applications could be integrated into the Results section and revised within the corresponding subsections. This applies in particular to the 3D geometry of low-angle normal faults, the geological implications of the DOMs and their comparison with seismic data, and the use of the dataset for monitoring deglaciation.
5. Further methodological aspects and points requiring clarification are provided in the line-by-line comments below. As these comments arise directly from the respective text passages, I have retained them in the technical comments section.
Technical comments and suggestions(!)
------------------
- Abstract: “new possibilities” may be overstated; method established for more then 10 years (see your own related-work references, e.g. Wilkinson et al., 2016).
L17: Unify number formatting (e.g. “12” vs. “twelve”). The number is also repeated within the same section.
- https://doi.org/10.5281/zenodo.18940068:
Consider creating one overarching Zenodo record for the complete dataset, with links to the individual subsets. This could avoid 12 separate references for what is presented as one dataset.
L35ff: “including low angle normal faults (LANF) systems found in present-day, deep-water, rifted margins in the South China Sea ...” — Clarify that the following items are examples in a list.
L47ff: “Geological investigations in this region typically rely on ‘conventional’ field methods and are strongly limited by accessibility.” — Specify what is meant by “conventional” field methods; consider combining with the preceding sentence.
L53: “the Err” — Check font style/formatting.
Figure 1: Add units to the x-, y-, and z-axes
Figure 2a: Red and purple are difficult to distinguish; consider using more contrasting colors.
L94: “fly-bys” — Check whether the hyphen is needed; “flybys” may be preferable.
L95: “while manually re-focusing was done in between” — This change the internal camera parameter ("focal length") and influnce the quality of reconstruction. Should be discussed
Table 1: Clarify the meaning of “Distance camera/outcrop normalized.” What does the scale represent, and what are its units?
L110: Instead of emphasizing the number of points in the text, consider reporting model-quality indicators, such as camera-position accuracy and/or the mean reprojection error, as these are more relevant for the resulting dense point cloud and DEM.
L110 and L116: Consider rounding the point-cloud sizes to millions, as only the order of magnitude is relevant here.
L121ff: Delete these lines.
L130: “Post-Processing Kinematic data” — Please clarify; do you mean PPK-corrected GNSS positions?
L131: “natural markers” — It is unclear how these markers and their coordinates were obtained. Were corresponding points manually selected in the SwissALTI3D DEM and the point cloud? Please explain.
General: The number of DOMs is repeated frequently (often “10”). Consider reducing these repetitions after stating it once in the relevant sections.
Table 3: Use consistent units (cm or m).
L159: Are “DOMs” or “DEMs” meant here? Please clarify, preferably in the Introduction, how “DOM” is defined in this study: only the textured 3D mesh, or the complete set of products including DEM, orthophoto, point cloud, and mesh?
L174: Maybe add “... overlap of the captured images”
L185: What is the implication of this variation? Does it affect the accuracy or spatial resolution of the results? Consider addressing this in the Discussion.
L213: “Quality was ensured through manual inspection and (internal) error analysis.” — Clarify what “internal error analysis” and “manual inspection” refer to. Does the former correspond to the error metrics listed above?
L214: “model-wide” and “localized DEM differences” — Please clarify these terms. Do you mean overall model differences and local DEM differences? The sentence is unclear.
Figure 3: Consider narrowing the elevation-difference scale. Since most differences lie between −1 and 1 m, would a narrower range provide more detail and make smaller spatial patterns easier to identify?
L216: Are these values really always zero? To assess whether the differences may partly result from georeferencing offsets, consider also reporting the mean difference, mean absolute difference, and/or median.
Figure 4: Add units to the axes.
L267: "... robust and stable surface features such as the CP4 in model 9 with an error in xyz of 0.21 m (Table B9)"
Consider focusing on more general patterns, comparisons between the derived parameters, and their relationships, as the detailed results have already been presented and are partly repetitive.
L270: The rationale and criteria for selecting the control points would be better placed in the Methods section.
L294: I am not convinced by this statement. Recent UAV–SfM studies using RTK-GNSS report differences on the centimetre rather than metre scale (e.g. https://doi.org/10.5194/isprs-annals-X-1-2024-169-2024).
L348: “This can be explained by the acquisition of 3 045 images and 13 flights to cover an area of 4.08 km² for model 11 while model 3 is only based on 1 132 images and 8 flights acquired over a much larger area of 9.84 km² (Table 1).”
I am confused by this explanation. The number of images and survey area alone do not determine spatial resolution or ground sampling distance, which mainly depend on sensor and pixel size, focal length, and flight height. Please clarify.
L354ff: The following aspects are difficult to follow because the corresponding observations are not clearly presented in the Results. Please clarify what is blurred—the orthomosaic or the DEM—and how this relates to image overlap. It is also unclear why a pre-planned autonomous mission would avoid these effects.
L361ff: As in the previous comment, the described effects are not clearly visible in the figures or presented in the Results. It is useful to mention them, but they should be described more clearly. Statements such as “The differences must be minor so they cannot be interpreted as geological structures or variations in lithology” appear speculative and should be better substantiated or rephrased.
L372: The statement is too vague in my opinion. Elevation estimates are also affected by image overlap and tie-point distribution. The main difference is that height determination is more sensitive to weak image geometry, such as nadir-only imagery, which can result in unfavorable intersection angles and increased sensitivity to calibration errors.
Section 5.2, “Comparison with other DOM datasets”: Most of these aspects might be better placed in the Introduction, as they mainly describe the state of the art, define the research gap, and motivate the aims of the dataset. The main points could then be briefly revisited in the Discussion or Conclusion.
L430: “All data were reprocessed following the protocol described above with 15 control points, whereas the georeferencing of the first model was based on GNSS data (total error of 0.25 m).” — Unnecessary repetition; consider shortening or deleting.
L515: Consider listing all published data types (images, GCPs, DOMs, DEMs, orthomosaics, etc.) near the beginning of the manuscript. Are the natural markers (image coordinates) also included in the dataset?
L525ff: The following paragraph is poorly formatted and repeats information already provided in Table 5.
L551: This appears redundant with L555; consider merging both statements.
L569: While I also hope this will be the case, the statement seems too informal for the manuscript and should be rephrased more formally.