the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 05 Mar 2026
UAV-based Ultra High-Resolution Geodetic Mass Change Estimations near Schirmacher Oasis in East Antarctica: Insights from Sub-seasonal, Seasonal and Annual Timescales
Abstract. This study presents an Uncrewed Aerial Vehicle (UAV) based framework to quantify surface elevation changes and subsequently geodetic mass change at high spatial and temporal resolution. The approach was demonstrated over a ~ 6.7 km2 area near the Schirmacher Oasis, Dronning Maud Land, East Antarctica. High-resolution imagery derived from UAV allowed application of different pixel-wise snow and ice densities for precise geodetic mass change calculations. Sub-seasonal mass change values showed a gain of 0.539 cm w.e. during 17 November–3 December 2023 and a loss of –0.144 cm w.e. during 3–17 December 2023, highlighting the influence of short-term meteorological drivers such as temperature fluctuations, snowfall, and snow drift leading to snow redistribution over the ice surface. Over the seasonal period (17 November–17 December 2023), the geodetic mass change was found to be –0.141 cm w.e., while annual estimates exhibited a gain of +2.072 cm w.e. for the surveyed area, and +0.751 cm w.e. for a larger coverage (3 December 2023–21 November 2024. We validated the elevation change with in-situ stake measurements, which showed very good alignment. We also discuss key operational challenges such as flying in extreme weather conditions, battery limitations, and geolocation issues- and offer practical recommendations to improve the reliability and scalability of UAV-based monitoring in polar regions. With the suggested recommendations, the demonstrated framework can be applied to other polar sites to enhance understanding of ice sheet surface processes, to develop sites for calibration and validation of satellite-derived geodetic mass change products.
- Preprint
(14060 KB) - Metadata XML
-
Supplement
(939 KB) - BibTeX
- EndNote
Status: open (until 05 Jun 2026)
- RC1: 'Comment on egusphere-2025-5601', Kriti Mukherjee, 27 May 2026 reply
-
RC2: 'Comment on egusphere-2025-5601', Anonymous Referee #2, 27 May 2026
reply
General comments
The authors have collected Uncrewed Aerial Vehicle (UAV) data over a blue ice area in Antarctica to estimate the geodetic mass balance at different temporal scales, ranging from two weeks to one year. To my knowledge, such a detailed study of mass balance in Antarctica has not been presented before, and in addition to a novel approach, the authors also interpret the results and provide practical recommendations for the use of UAVs in harsh environments. I do believe that the study objective (a high-resolution estimate of mass balance over a blue ice area) contributes to understanding the inherent variability between and within these unique areas much better. This objective has important implications for surface mass balance modelling (for example, how can results be downscaled using surface topography), as well as on survey design (for example, at what spacing do ablation stakes need to be placed to capture local variations). However, I have some major concerns regarding the presented results:
- The authors use only three stakes for validating the results (in the large area, in the small area it seems like only one stake overlaps). The stakes are not representative for the full survey area (only in ice) and show agreement in only two of the three stakes.
- The sub-seasonal values do not add up to the seasonal value. The reverse patterns in the sub-seasonal values suggest artefacts resulting from the DEM generating process. The authors do not report the number of GCPs used. The reported errors (no indication of how many points were used to estimate those), notably in vertical direction (Table 2), exceed the average observed elevation changes.
- The authors assume that measured elevation changes are representative of “local mass change signals”. However, the presence of blue ice is often associated with complex flow patterns, and the assumption of a minimal contribution of ice flux divergence (or convergence) is not supported by observations (even though these are likely available through the stake network and/or other GCPs). The stake readings in combination with GPS measurements could be even used to translate the observed elevation changes to surface mass balance estimates. Of course, this is only feasible if there is a reasonable number of measured stakes.
These doubts limit confidence in the robustness of the results, and interpretations regarding observed patterns risk being influenced by artefacts, noise, or assumptions. A more transparent reporting of the number of stakes and GCPs could partially increase the credibility, but if these numbers are small, the study should focus on the methodology instead of on interpreting patterns. An alternative would be to extend the time series and show that patterns are actually consistent but given that Antarctic fieldwork is involved this might not be feasible.
I have provided more detailed doubts and suggestions below.
Specific comments per section
Abstract: It would be informative if numbers are mentioned with range over the area and/or uncertainty, and if the extent and variability of snow cover are discussed.
Line 12: Sub-seasonal values do not add up to seasonal values.
Line 16: what is the extent of the larger area?
Line 17: how many in-situ stake measurements? And how can you validate elevation change with in-situ stake measurements? Do you consider only the surface mass balance, or do you also consider the ice flux?
Line 17: The alignment is not good, with a large difference (~0.15 m) in the Vettaiya stake, beyond your estimated uncertainty (Figure 7).
1 Introduction:
Line 32: “Studies suggest one of the major positive drivers” -> “The major positive contribution to Antarctica’s SMB is snowfall”
Line 34: “plays a crucial role” -> “can play a crucial role”
Line 36: How would SMB stabilize the ice sheet? Needs a reference
Line 38: This statement needs some refinement. Maybe something in the line of: “capturing local spatial variability is challenging give the spatial resolution of satellite or model data”. In other words, it needs to be specified what limitations are meant.
Line 41: Maybe start a new paragraph about UAVs, previous paragraph is about SMB and traditional measurement techniques.
Line 41: what does “it” refer to?
Line 41: “which are commonly known as drones” -> remove
Line 43: offering -> offer
Line 43: Compared to stakes, there is not an improvement in terms of temporal coverage, there remain logistical constraints, there is a very limited spatial coverage. So definitely agree with that it acts as intermediate platform between satellite and ground-based measurements, but needs to be introduced carefully. Highlight only the limitations of other techniques that can actually be overcome with UAVs.
Line 51: Some references seem to be missing?
Line 46-56: Very elaborate introduction about UAVs, can be summarized in one or two sentences to make the text more concise.
Line 60: how about mountain glaciers? There are extensive mappings of, for example, the Morteratsch glacier in Switzerland, over time spans of multiple years. This claim is rather bold, and this study itself also focusses on a small area of less than 7 km2. At least define what are “small areas”, and specify whether UAV-based geodetic approaches are considered or only the ones in Antarctica.
Line 63: Importantly -> To our knowledge, no previous studies..
Line 65: which gaps?
Line 66: section --> small area (~7 km2)
Line 67: I think stake measurements measure the ablation/accumulation, the actual SMB. Geodetic mass change estimates need to be compared to measurements of actual change of the surface height and density to demonstrate accuracy.
2 Study Area and Fieldwork:
Line 77: do you know how strong?
Line 83: is this blue ice area a result of the enhance surface ablation processes? Maybe more intuitive to write “enhanced surface ablation processes, … areas of cDML, resulting in extensive exposure of blue ice”?
Line 83: The sentence about the runway feels out of place, although the following sentence hints that the authors do this research in part to understand the durability of the runway and/or to help with its maintenance I think. As this is not further addressed in the study, I would suggest to remove it.
Figure 1: Maybe nice to show the entire continent in the first panel, if possible, in the same colors as the second panel (the latter is more for aesthetics).
Line 99: I do not understand why the authors use a base station that is not actually a base station. How did they mark the location for remeasurements? Using a fixed point as reference is essential for understanding the dynamics between two measurement times. In my opinion unnecessary errors are introduced by using the Emlid base for PKK correction – why not use the base station at Maitri Station directly for this? The baseline seems to be roughly 5 km, which is a very normal distance. I am interested to see the difference in results between processing with the Maitri Station and the current approach. Of course this would only be possible if the base station at Maitri Station is set up at exactly the same spot for all measurements, for example marked by a nail in the rocks. If this is not the case, the biases in positioning are likely higher than the signal observed. In the horizontal plane, this can be partially overcome by using the very small coverage of exposed bedrock for co-registration in the larger area datasets. In the vertical plane this is also possible, but will only rely on the results of the triangulation. The fact that this bedrock is only in one corner of the imagery makes it more challenging to use it as a valid constraint.
Line 112: I think there is REMA data at 2-m resolution in that area (see https://www.pgc.umn.edu/data/rema/), which should provide a reasonable accuracy for following terrain features at a flying height of 260 meters. Maybe just mention that “All flights were executed at a fixed altitude relative to the take-off point.”
Line 106: What was the battery duration? (Actual vs factory specifications)?
Line 107: at what altitude?
Line 115: initial trial experience: that then refers to the generation of DEMs and orthomosaics?
Line 116: Three flights per day? Or three days of coverage?
Line 117: Although it is very useful, the experience and recommendations would fit better in a field report than in a scientific study. Would suggest to refer to a field report where these are documented or to include them in the supplementary materials.
Line 121: why is the data of 3 December chosen for the annual window and not one of the other dates in 2023
Table 1: Is the overlap in lateral or frontal?
Table 1: The information in the table is rather repetitive. It would guide the reader more to have a scheme that explains what is subseasonal, seasonal, and annual. All other information in the table can be summarized (number of flights, total number of images, elevation). If the authors decide to keep the Table as it is, I wonder whether PPK corrected and standalone can be combined into a single column? And are the standalone images used in the processing? If not, I would just report the PKK corrected number of images and remove the other two columns.
Line 128: traditional in-situ stake measurements do measure the surface mass balance, so how can these be used to validate a geodetic mass balance?
Line 130: How many stakes are in the stake network?
Line 131: How deep are these stakes in the ice?
Line 132: GCP locations: are these GPS locations?
Line 134: How was this measured?
Line 137: what do you mean by elevation change? The height of the stake above the surface? If so, this conversion is just to change the units of the SMB, and not referring to the geodetic mass change.
Line 144: normally wind chill is calculated based on the temperature and the wind speed, and it is not measured/collected by the AWS.
Line 145: Where is this measurement gauge positioned?
3 Methods:
Line 155: high-performance?
Line 160: where are the GCPs? How many are there? What are check points? How many are there? Including them in one of the figures would be helpful. I see that this is explained later in the text. Would help the reader to make sure these terms are introduced earlier. The number of points need to be specified.
Line 169: for how long were the GCPs measured in the field? And how was the GNSS set-up? With a tripod or is it a handheld device? What are the accuracies of the vertical coordinates?
Line 174: what is mildly filtering? What kind of filter and what parameters are used?
Line 180: how is this accuracy in Table 2 calculated? Is it something like a MSE? The errors reported in the Table exceed the observed elevation changes by a factor of 10 or more, which makes it difficult to assess the robustness of the presented elevation-change estimates.
Figure 2: what does “(right)” mean in the caption? Panel a and b would need a scalebar.
Line 194: What is “leading to a melting scenario”? The authors assume that if the surface transitions from snow to ice, and has a negative elevation change, that all the snow is removed. It is indeed a negative surface mass balance, and likely that this happens, although it is not excluded that besides snow also some of the ice erodes (I find the assumption fair though). The surface lowering could be a result of the ice flow, sublimation, ablation, melt, or abrasion or snow drift. The latter could be likely, i.e., that snow dunes are removed by accelerating winds. It would be interesting to analyze what happens in the areas that transition from snow to ice to validate the results. It is rather unlikely that the elevation increases in these areas, if this is the case either there are some interesting ice dynamics, or it indicates that the changes are smaller than the uncertainty.
Line 196: how can an ice surface experience accumulation? The only scenario I can imagine is some kind of freeze on of meltwater or rain, but that seems rather unlikely. So the ice surface raises in between two captions, it is a result of flow dynamics or it indicates some artefacts resulting from the data processing. Can the authors quantify how much ice surface is lifting?
Line 203: What does “sufficiently consistent with the evolving surface conditions over the analysed region” mean?
Line 206: Even if the ice flow velocity is low, the ice flux divergence can have a significant influence on elevation changes (especially considering the very small signal that is detected). Only if the ice flow velocity is uniform over the region these influences are less. A motivation for assuming that elevation changes are representative of local mass change signals, would be the comparison between stake measurements and surface elevation measurements (GPS based for instance). If these two correspond, the ice flux divergence is indeed not contributing. Another option would maybe be to calibrate the observed DEM differences directly with the stake readings. But in both cases, a sufficient amount of stake readings is necessary to rule out any outliers, especially because the temporal resolution is high.
Line 208: This masking does not appear in Figure 3, where the roads are visible. Is the masking applied nonetheless?
Line 233: So the uncertainty of the geodetic mass change depends on the actual volume change? Is that realistic? (Assuming that Delta is Sigma in eq 8).
4 Results:
Line 239: How many checkpoints?
Line 240: How is the accuracy calculated?
Line 242: Before it is mentioned that this process is done only for the ground checkpoints, but here it is mentioned that it is both for the ground control points and check points. What is finally reported in Table 2?
Line 256: I think the spatial distribution is for each date of data collection, not for each interval.
Line 257: Figure S3.
Line 258: interval should be date
Line 261: interval should be date (e.g., sub-seasonal should be 17 nov 2023)
Line 269: statistically significant?
Line 266: This range of values seems very large over a timespan of two weeks. It is remarkable that the range of values for short periods is similar to the range of values for the annual period (Table 3). I wonder whether this just implies that the observations are noisy.
Line 269: lowering? or increase? Lowering is negative and increasing is positive, right? On average there is a elevation increase, which is not intuitive given that more ice becomes exposed.
Line 271: this contrasting pattern could hint at artefacts in the second DEM.
Line 279: Did the authors observe such extreme snow drift in the field? For example around the skidoos or on the roads? I would expect the exposure of ice to change drastically if this were the case, while the exposure of ice seems very similar in the different periods.
Figure 3: An elevation increase of +20 cm over blue ice (in the left center of panel a) in a time period of two weeks seems rather unlikely. How could this physically happen if there is no snow accumulation (according to Figure S3)? Drawing the outlines of the snow patches over this data could help in interpretation. A uniform colorbar centered around 0 for Figures 3, 4, and 5 would be helpful in interpretation.
Figure 4: Do the patterns correspond to migrating snow dunes? Is that visible in the imagery? Does the orientation align with the wind direction?
Figure 5a: Is there a map of blue ice vs snow for the first annual acquisition over the large area? Also, I am surprised that the top left corner, which seemingly covers exposed bedrock (Figure 1) shows a large elevation loss. Is there a physical explanation?
Figure 5c: I’m surprised with the elevation increase between 470 and 480 m elevation, which corresponds to an area that is first snow covered and then becomes exposed ice. This implies a surface lowering. Extent of ice for the two acquisitions could be indicated to help interpretation. Are the signs inverted?
Table 3: What is the error based on?
Figure 6: Why is the seasonal GMC not a sum of the two subseasonal values?
Line 339: In Table 4 the reported uncertainty of the UAV DEM is 0.01m
Line 342: “which are within the combined uncertainties (Fig.7)”. Figure 7 does not seem to support this statement.
Line 344: “shows close agreement across all three stakes”. There is agreement for two of the three stakes (within uncertainty range), but the third stake (Vettaiya) shows a rather large disagreement (~50%).
5 Discussions:
Line 364: If snow cover drops, I would expect a surface lowering, or mass loss. The results show the opposite, even while the authors argue that the main contribution to the geodetic mass change is the ice.
Line 371: Numbers of snow/ice coverage here do not correspond with numbers in Line 362-364 (or are they averages?), and other numbers are reported in Table 5.
Table 5: The snow and ice area should be reported for each date, and not for each interval. If I understand correctly, the authors use the four different scenarios for estimating the geodetic mass change (snow -> snow, ice -> snow, snow -> ice, ice -> ice). I do not understand why there are only two areas reported here if 4 are used for the conversion.
Line 398 – Line 547: This a very detailed interpretation of data with large uncertainties. The conclusions are speculative, unless my earlier expressed concerns with the estimated elevation changes could be resolved/clarified somehow.
Line 549 – 576: Relevant and useful for other studies, but would be better in supplementary materials or field report (see earlier remark)
Conclusion:
Line 605: remove “reliable”
Line 610: “our results aligned very well, showing its reliability.” The agreement appears variable across stakes, and the reliability appears limited given the low number of stakes for validation.
Appendices:
Figure S3: I presume the titles of the subpanels are wrong, and they should refer to the dates? The resolution is too low to read the labels of the elevation contours. It would be useful to see the actual drone imagery also instead of only the derived surface classification. Like in Figure 2, but then for every timestamp.
Citation: https://doi.org/10.5194/egusphere-2025-5601-RC2 -
RC3: 'Comment on egusphere-2025-5601', Anonymous Referee #3, 03 Jun 2026
reply
This manuscript is built on a potentially useful dataset from a difficult East Antarctic field setting, and repeated UAV surveys at this site could make a worthwhile contribution. My main reservation is not the ambition of the study, but whether the analytical chain is sufficiently documented and internally consistent to support the precision of the reported results. In its current form, I do not think the manuscript meets that standard. Several of the conclusions depend on centimetre-scale mean changes that are close to the stated uncertainty, the DEM reference-frame strategy is not yet convincing, and some of the reported numerical conversions appear inconsistent.
principal issues
At present, the most serious issue is internal consistency. Table 4 appears not to be self-consistent. Under the authors’ stated assumptions, the stake sites remained snow-free and point-scale mass change should therefore be approximately calculated from Δh × 0.917 in m w.e. That relation is broadly satisfied for the first row, but it does not hold for the Sankalp and AWS rows. For example, a height change of −0.306 m should yield roughly −0.281 m w.e., not −0.092 m w.e.; similarly, −0.205 m should yield roughly −0.188 m w.e., not −0.061 m w.e. The same discrepancy is visible in the UAV-derived column.
This is not a minor presentation problem. It raises concern that either the conversion workflow is incorrect or the method has not been described completely. I also note several contradictions between the text and tables, including the stated UAV DEM uncertainty versus the uncertainty listed in Table 4, as well as inconsistencies in the interpretation of snow/ice contributions and the sign of the reported seasonal and annual changes.
The manuscript indicates that stake positions on the ice surface were used for co-registration of the DEMs. This is a major concern. Stakes on moving ice are not stable reference objects, especially over an annual interval. If such points are used to align successive DEMs, the co-registration may either suppress real surface change or introduce spurious change. It also weakens the independence of the validation if the same stake network underlies both alignment and comparison.
The paper needs to state explicitly which points were used as GCPs, which were used only as checkpoints, and which were used solely for comparison.
The reported mean elevation changes for the two sub-seasonal windows and the seasonal window are only +0.004 m, +0.012 m, and +0.007 m, whereas the checkpoint vertical errors reported in Table 2 are substantially larger. Area averaging can reduce random noise, but only after systematic DEM-to-DEM bias has been shown to be negligible. At present, the paper relies mainly on checkpoint residuals and an adopted decorrelation length, but it does not convincingly show that residual bias after DEM differencing is small relative to the reported mean signal.
This is especially important because the manuscript’s main process interpretation rests on a reversal between the two short survey windows.
A substantial part of the manuscript’s claimed contribution rests on assigning density according to snow/ice class transitions. However, the actual classification procedure is described only in very broad terms. The manuscript does not specify the classifier, the training strategy, the basis for separating classes, or the classification accuracy. It also does not explain how ambiguous cases were treated, such as shaded surfaces, patchy snow over ice, wind-packed snow, or illumination differences between acquisitions. The density model is also highly simplified. Only two fixed densities are used, with no local density measurements and no treatment of intermediate surface states.
At some points the manuscript correctly states that ice-flux corrections were not applied and that the results should therefore be interpreted as elevation-derived mass change rather than flux-corrected mass balance. Elsewhere, however, the text reverts to “mass balance,” “surface mass balance,” and “geodetic mass balance” in ways that blur that distinction. This matters because, without a dynamic correction, the reported quantity is not equivalent to a full mass-balance estimate.
The validation is restricted to three stake locations, evaluated only at the annual timescale, and all located on snow-free ice surfaces. That is a narrow test relative to the mapped domain, particularly when the density-based method is most relevant for snow-covered terrain. This is also difficult to reconcile with the fact that all three stakes show local surface lowering, whereas the annual area-mean result is positive. That pattern may be real, but it means the stake comparison cannot be treated as broad validation of the spatially distributed results.
The slope, aspect, and meteorological sections are plausible and potentially interesting, but at present they read as more conclusive than the analysis allows. These sections are mostly descriptive. They do not include statistical testing, uncertainty by class, or a quantitative framework linking meteorological conditions to the mapped change fields.
Citation: https://doi.org/10.5194/egusphere-2025-5601-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 955 | 495 | 78 | 1,528 | 191 | 62 | 91 |
- HTML: 955
- PDF: 495
- XML: 78
- Total: 1,528
- Supplement: 191
- BibTeX: 62
- EndNote: 91
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Major Scientific Concerns
The manuscript does not sufficiently articulate the scientific motivation and broader relevance of adopting a UAV-based approach for geodetic mass balance estimation in Antarctica. While the technical workflow is clearly presented, the fundamental justification for using UAVs over established satellite-based methods remains unclear. In its current form, the argument is primarily introduced in the conclusion, where UAV observations are described as a means to validate satellite-derived mass balance. This reasoning is not fully convincing, as both UAV and satellite approaches rely on similar remote sensing and photogrammetric principles, differing mainly in spatial resolution and scale of observation. Satellite-based elevation change products, such as those derived from altimetry and DEM differencing, are already well-established, widely validated, and routinely used in glaciological studies. Therefore, the authors should more clearly explain—preferably in the introduction—what specific scientific or methodological gap their study addresses. In particular, it would strengthen the manuscript to explicitly demonstrate whether UAVs are intended to resolve fine-scale surface heterogeneity, capture short-term processes, or provide calibration data that cannot otherwise be obtained, and why such capabilities are essential for advancing current knowledge.
The terminology used to describe temporal scales in the study also requires clarification. In glaciology, mass balance is typically defined over seasonal periods (summer and winter), each spanning several months, often between four and eight months depending on the region. In contrast, the study refers to intervals of approximately two weeks as “sub-seasonal” and a one-month period as “seasonal,” which may be inconsistent with standard terminology. This usage risks confusion and limits comparability with existing studies. The authors should clearly define these terms and justify their application, particularly if they are intended to represent operational survey windows rather than climatological seasons. If this terminology reflects specific Antarctic field practices, this should be explicitly stated. More importantly, the manuscript should better explain the scientific relevance of such short-term mass balance estimates and how they contribute to a broader understanding of surface mass balance processes.
A further limitation of the manuscript lies in its relatively weak engagement with existing literature on UAV-based glaciological applications. While the authors note that previous studies are spatially limited and focus on short time periods, these limitations also apply to the present work, which covers a relatively small study area and primarily a one-year interval. As a result, the novelty and contribution of the study are not sufficiently contextualized. The manuscript would benefit from a more comprehensive review of prior UAV-based mass balance studies, both within Antarctica and in other glaciated environments such as the Arctic, the Himalaya, or alpine regions. In cases where Antarctic studies are scarce, comparisons with analogous studies from other regions would provide valuable context. A more critical discussion of the state of the art—highlighting what has already been achieved and what challenges remain—would allow readers to more clearly assess the contribution of the present work.
Related to this, the manuscript does not clearly synthesise its key contributions. Although elements such as high-resolution mapping, temporal variability, and pixel-wise density estimation are presented, they are not explicitly framed as distinct advancements relative to existing approaches. Given the limited spatial and temporal extent of the study, it is important for the authors to clearly articulate what is novel, whether this lies in methodological innovation, improved process understanding, or the integration of datasets at different scales.
The proposed density assignment framework, based on snow–ice transitions and elevation change, is a potentially valuable contribution, but it requires stronger justification and discussion. In particular, the manuscript should more explicitly compare this approach to traditional methods that use constant density assumptions or firn densification models. The uncertainties associated with applying fixed density values (e.g., 300 kg/m³ for snow and 917 kg/m³ for ice) should also be discussed in greater depth, especially given the known spatial and temporal variability of snow and firn properties. Any limitations arising from the lack of in situ density measurements should be acknowledged more explicitly.
Additionally, the study remains highly localised, and its implications for larger-scale Antarctic processes are not sufficiently discussed. While small-area studies are valuable for process-level understanding, the manuscript would benefit from a clearer explanation of how these findings relate to regional or ice-sheet-scale surface mass balance patterns, or how they might inform the interpretation and validation of satellite observations. Without this broader linkage, the significance of the results remains somewhat limited in scope.
The discussion of uncertainties, although present, could also be expanded. Specifically, it would be useful to assess the relative contributions of DEM errors, classification uncertainties, and density assumptions to the overall mass balance uncertainty. A comparison with uncertainty levels typically associated with satellite-derived mass balance products would also help to place the UAV-based approach into context.
Finally, the presentation of supplementary material requires improvement. The supplementary figures are referenced inconsistently and not always in a logical sequence within the main text, which makes it difficult for the reader to follow the analysis. A more structured and sequential referencing of supplementary material, aligned with the flow of the main figures, would significantly improve readability.
Overall, while the study demonstrates a technically robust application of UAV-based photogrammetry for mass balance estimation, its scientific contribution would be substantially strengthened by clearer motivation, improved contextualisation within existing literature, more rigorous discussion of methodological choices, and a stronger linkage to broader glaciological questions.
Line-wise and Section-wise Comments
Abstract: The loss values need to be accompanied by corresponding uncertainties. The term surveyed area and larger coverage are misleading. Isn’t the larger coverage surveyed? Why doesn’t the annual period start from 17th November 2023?
L29: can you provide the value in m in terms of sea level rise?
L56: The author’s names should not be in brackets, just the year.
L63: Do you mean for AIS? For high mountain glaciers there are many studies using UAVs. Please correct or be specific.
L121: Again, why is the annual window not 17 November 2023-21 November 2024?
Table 1: What is the unit for standalone and what does it mean?
L135-139: Please revise the sentence avoiding repetitions. Please mention a reference for the ice density value you used for mass balance calculations.
L154: Multi Stereo View -> Multi-View Stereo
L175: Please put a comma after sub-seasonal
Figure 2: It is not clear what is meant by ‘(right)’ in the figure caption. Please reduce the number of ticks for the grid lines or use decimal degrees for clarity.
L220: Please revise the sentence avoiding repetitions.
L237: Please delete ‘Generation’ from the heading.
L257: I don’t see Figure 3e. What is Fig S3a-b etc.? Supplementary information? I think you should refer them appropriately. I don’t see any references to S1, S2 etc. If you are not referring to anything in the main text, they should not be in supplement too.
L268: I think this is left-skewed – longer tail to the left.
L269: Not clear what you mean. Please revise the sentence.
L274: This is right skewed – longer tail to the right.
L288: What is neutral distribution?
Figure 6: It is strange that the annual mass balance is so different for the larger area while the thickness change patterns look similar in 5a and 5c. What does this suggest?
L340-342 and Figure 7: As per my judgement, AWS stake measurement is outside the uncertainty range of UAV measurement.
L407: Please remove the lines to represent absolute value. Geod Mass Change -> geodetic mass change.
L398 - Section 5.3: These two sections have too many details. These details would be useful if this is a general phenomenon over a larger area or over multiple years. As your analysis is only valid for a very short period, you need to establish the relevance of these details to the glaciological community. Please correct the numbering of the subsections.
L512 – Section 5.3: Please correct the numbering.
L514: delete ‘change’