Summary and overall assessment

The paper presents the first systematic field evaluation of dock-based UAV systems for geohazard monitoring across three alpine environments and introduces an automated end-to-end workflow for displacement and change detection. The topic is timely and important; however, several sections remain too general, and the case-study specific results and methodological details need expansion to demonstrate what was concretely achieved and learned at each site

Here are the major comments. Further comments are in the pdf.

• Clarify what dock-based UAVs enabled beyond multitemporal UAV campaigns
• The Paper should explicitly state how the dock-based approach changed feasibility, frequency, reliability, and safety compared to conventional campaign-based UAV surveys, and tie this to site-specific outcomes. Add for each site, a brief paragraph that contrasts “what would have been possible with manual UAV campaigns” vs. “what the dock enabled” . The advantages and limitations of the method should be discussed.
• Define a clear research question per case study with geohazard focus
• Expand the Methods section that the reader can follow the result chapter. Describe the workflow steps in more detail and explain the differences between multi-temporal UAVs. Specify which datasets/timings were compared. E.g. whether successive pairs (A–B, B–C, C–D) are processed, or whether all time points are referenced to a baseline (e.g., A–B, A–C, A–D), and why (e.g., to maximize sensitivity to acceleration vs. minimize decorrelation). This is important to understand the results.
• Accuracy assessment: Whether a detailed accuracy assessment is necessary depends on the research question. When analysing small deformations close to the detection threshold, such as those at the Skjold rock slope, it is crucial to understand the limitations of dock-based UAVs. A level of detection is necessary for all study sites (even for every data pair) when analysing displacements.    Consideration of accuracy is important for the applicability of dock-based UAVs for continuous deformation monitoring.
• The result section should focus on the differences between multi-temporal UAV and dock-based UAV and its significance to geo-hazard monitoring.
• Restructure: move general dock-based context from Discussion to Introduction; focus Discussion on evaluating the presented method and results.
• Add a subsection with advantages and limitations.

This is a review of the manuscript by Maschler et al., entitled “Automated UAV systems for geohazard monitoring: case studies from the Supphellebreen icefall (Norway), the Skjøld instability (Norway), and the Blatten landslide (Switzerland)”, submitted for publication in NHESS. The authors present the application of a commercial dock-based Unoccupied Aircraft System (UAS), together with a new integrated data processing pipeline, for persistent geohazard monitoring in mountainous terrain, in the context of three different case studies. Once installed, state-of-the-art dock-based UASs facilitate quasi-autonomous photogrammetric surveys (and other types of measurements) on demand or at regular intervals, even beyond visual line of sight. Hence, they have great potential to complement ground-based and spaceborne systems for geohazard monitoring and risk assessment, especially in complex alpine terrain. Dock-based UASs might be of particular interest in cases where, for example, the study area is difficult to access, located within a risk zone, or beyond visual line of sight, and where frequent measurements or long-term monitoring are necessary. However, dock-based UASs come with certain practical limitations, as they are relatively costly (compared to conventional campaign-based UAV surveys), require a stable power supply, and must conform to regulatory and legislative frameworks, which may prevent timely installation and operation. The manuscript is easy to read and fits the scope of NHESS. I acknowledge the effort the authors put into the setup of the dock-based system, the data collection, and the development of the data processing pipeline. However, I fully agree with the other reviewer that a major revision is necessary to specify the aim of the study, emphasise the added value and limitations of the presented monitoring approach, and outline which broader conclusions can be drawn from the three different case studies.

I have attempted to complement Reviewer 1’s report, though some overlap is inevitable given the shared main concerns.

General comments

1. Specify the main aim of the study (see also Review 1) in the abstract and in the last paragraph of the introduction. Briefly discuss the benefits and limitations of different geohazard monitoring techniques (e.g. ground-based, spaceborne, airborne, conventional campaign-based UAV surveys) in the introduction, and emphasise in which cases a dock-based UAS might be a game changer (e.g. financially, technically, practically, research-wise) and in which situations multi-temporal UAV surveys would simply do the job. The study of Walter et al. (2022) provides a nice example of a dock-based UAS for repeated catchment-wide mapping of sediment dynamics and the monitoring of a debris flow torrent in alpine terrain. You cite their paper, but you could outline in more detail how your monitoring approach adds and relates to their and other studies.

2. The motivation for the selection of the three specific study sites is vague. What were the criteria for the selection? The apparent link between the case studies is the different geohazards (glacier lake outburst flood, rockfall, ice-rock-debris avalanche), but a bit more context on the magnitude and impact of previous glacier lake outburst floods at Supphellebreen and the rockfall activity at Skjøld would be helpful to understand how critical continuous monitoring at these sites is. Moreover, please also be more specific about how the frequent UAV surveys aided “ongoing hazard assessment and response efforts” in Blatten. It seems that operational monitoring with the dock-based UAS was not possible in Norway because of aviation safety regulations. Did you try to obtain permission for operational use, or was the test setup planned as such right from the beginning (see general comments below)? Please clearly state the aim and purpose of each case study and explain how they complement each other and contribute to the overall aim of the manuscript.

3. Since this manuscript may become a basic reference for future studies dealing with autonomous UAV surveys in complex terrain, careful use of terminology is essential. I am wondering why the authors refrained from using or discussing the adjective “autonomous” and instead use the description “automated UAV systems” when referring to their dock-based UAS. An “automated UAV system”, as stated in the title, could be any (off-the-shelf or customised) UAV that follows a pre-determined flight route and is only controlled by a pilot in the case of unforeseen events. However, from my point of view, the manuscript focuses on something different: the application of an autonomous UAS in alpine terrain, although full operation might not (yet) be possible in all cases due to regulatory and legal frameworks. Careful differentiation between automated, automatic, and autonomous UAVs is important, as different operational restrictions may apply (see e.g. https://www.easa.europa.eu/en/faq/116449). In this context, I would appreciate it if the authors could provide additional information on how much human intervention was necessary during the surveys and subsequent data processing. For example, was a remote pilot in a control room mandatory for continuous flight monitoring at Blatten? Was the data processing initiated automatically after upload from the dock, or was this done manually?

4. Since the authors identified current regulatory frameworks as one of the main barriers to the (operational) application of autonomous and beyond visual line of sight (BVLOS) flights, the legal requirements and undertaken administrative steps should be briefly discussed for each site and outlined in more detail in the general discussion section.

5. The advantages of the presented geohazard monitoring approach are mentioned at several locations in the manuscript, which is fine, but the manuscript would benefit if the limitations and challenges were discussed equally thoroughly. The following aspects, for example, would be worth addressing: current regulations and required permits for autonomous and BVLOS flights (potential conflicts with on-demand operations), safety measures, power supply of the dock-based UAS, data transfer, and UAS maintenance, especially in remote mountain areas.

6. Always define acronyms before using them for the first time (e.g. GNSS, InSAR, RTK, PPK, GCP, NVE, SORA, NTRIP, AOI, FH2), even if they are widely used, and keep their number to a minimum so that non-experts can also follow the manuscript easily.

7. Are there any restrictions or why are the code and data not published along with the manuscript? That is against common practice. Please have a look at the NHESS Data policy: https://www.natural-hazards-and-earth-system-sciences.net/policies/data_policy.html

Specific comments

Line 1-3: I suggest rephrasing the title. Consider including “autonomous” and/or “dock-based” as well as “mountainous/alpine/complex terrain”. It is probably not necessary to mention the three study sites explicitly. In any case, I would replace “landslide” with “rock-ice avalanche”.

Line 15-29: Specify the aim of the study, explain the motivation for selecting the three specific study sites, and briefly discuss both the potential and the limitations of the presented monitoring approach.

Line 35: “develop into multi-hazard cascades” → Please specify what you mean or provide an example.

Line 37-38: “Monitoring is often the only way to predict hazardous events” → Please modify this statement, as modelling and knowledge from previous events are also important.

Line 43: “To overcome these challenges” → Perhaps rather compare ground-based (stationary) and mobile measurement techniques, as differential GNSS, photogrammetry, LiDAR, etc. can be deployed within both ground-based systems and UAV/airborne systems.

Line 47: “Yet, conventional UAV surveys remain largely manual and campaign-based, which is often limiting their frequency” → I disagree. Conventional UAV surveys are usually campaign-based, yes, but not necessarily performed manually. Comprehensive surveys are usually performed using automatic UAVs. Please be careful with the terminology (see general comment).

Line 49-50: “as they reduce the need for human intervention” → Please specify what you mean by “human intervention”. On-site visits might be limited to installation and maintenance of the system, but a remote pilot is usually required for flight monitoring and safety reasons.

Line 52: Please be more precise. The paper by Walter et al. (2022) deals with monitoring sediment dynamics in the context of debris flow torrents, so it also has a geohazard focus, similar to this study.

Line 55: “systematic field evaluation of a drone dock-based automated UAV system” → Please clarify what you mean by “systematic evaluation”. Outline how it adds to or relates to the above-mentioned study by Walter et al. (2022).

Line 86: “These three locations, chosen for their contrasting mass-movement characteristics and monitoring challenges” → Please state this already in the introduction and explain in what sense the mass-movement characteristics and monitoring challenges differ.

Line 97: You state that no previous studies exist, but below you cite several studies with respect to Supphellebreen. Please clarify.

Line 98: “Temporal (diurnal and seasonal) variations…” → How do you know this? Please provide supporting evidence or a reference.

Line 103: This information is needed right from the beginning to understand the motivation for selecting this site. What were the characteristics, magnitudes, and impacts of previous GLOFs?

Line 105: I think the reference “Breien, 2008” is missing in the bibliography. Please check.

Line 120: How do you know this? Can you provide an appropriate reference? A bit more background on the geology would be helpful. How exceptional or critical is this site risk-wise, considering the presumably large number of slope instabilities in Norway?

Line 131: “that partially impacted local infrastructure” → To what extent? Please be more specific.

Line 150: It is unclear at this stage whether you are using a ready-to-operate commercial system or presenting a newly developed dock-based UAS. Clearly state which individual components (hardware and software) were implemented and which were newly developed.

Line 163: Can you provide a rough cost estimate for the dock and UAS?

Line 164-165: Some information on the power supply and consumption would be helpful.

Line 178-180: I assume the flights were conducted autonomously and BVLOS (and above 120 m), so a Specific Operation Risk Assessment (SORA) was necessary (https://www.easa.europa.eu/en/domains/drones-air-mobility/operating-drone/specific-category-civil-drones/specific-operations-risk-assessment-sora). Can you provide information on the legal requirements and how a permit was obtained within such a short time?

Line 201: Is the data processing initiated automatically after upload, or is this done manually? (see general comment)

Line 205-206: Could you explain why you use two different photogrammetry software packages (DJI Terra and OpenDroneMap) to create the point clouds, orthophotos, and digital elevation models? I am not familiar with DJI Terra, but in OpenDroneMap there are many processing options. Please specify whether default settings or any specific settings were applied. You refer to the ODM GitHub repository as a reference, so I assume the access date should be 2025 and not 2020. Two other suitable references are Toffanin (2019), a user guide from the core developer, and Groos et al. (2019), the first case study using OpenDroneMap for UAV-based photogrammetry in alpine terrain.

Line 238/250: Please describe your observations here and leave the conclusions for the discussion section.

Line 240-248: “Monitoring of the icefall at Supphellebreen” → The accuracy and level of detail of the velocity and surface change products are impressive, but it is unclear how these observations are linked to the UAV-based monitoring and risk assessment of GLOFs at this site, which was reported earlier in the manuscript as a motivation for this study.

Line 250-260: Given the number of flights at this site (n = 7), what is the added value of the dock-based system here?

Line 264-265: “The automated monitoring campaign following the rock-ice avalanche event at Blatten highlights the possibilities offered by automated UAV systems in post-disaster response” → Please specify this and provide examples.

Line 267: “could only be visited once” → Once during the entire monitoring period?

Line 299: “Automated UAV systems offer a paradigm shift in the monitoring of dynamic geohazard environments” → A paradigm shift in what sense? This is a strong statement and should be supported by sufficient evidence and robust arguments.

Line 341: “4D datasets offer a robust foundation for training deep learning models” → Please be more specific or provide an example.

Line 356-358: Is this not a bit too optimistic? During precipitation, strong winds, and ground fog, UAV surveys are usually not feasible or deliver data of reduced quality. Can you quantify the number of days with stratus clouds where the base height is high enough for UAV surveys to be performed safely?

Line 357: “low fog” → Do you mean ground fog or stratus clouds (i.e. low-level clouds)?

Line 365: Why a maximum of 120 m above ground level? Do you refer to the maximum flight altitude in the “open” category defined in the EU drone regulation?

Line 374: Please provide a rough cost estimate here.

Line 394: “failure event” → Consider using “rock-ice avalanche” or “rockfall and glacier collapse”.

Line 397-398: Have autonomous surveys up to 3000 m a.s.l. been performed? Can you explain what “dynamic mission adjustment” meant in practice? Both information should be included in the methods section.

Table 1: Could be moved to the appendix. I think an overview table with key figures regarding the dock and UAV would be more interesting.

Table 2: Maybe add the maximum distance of the UAV from the dock during the surveys.

Figures 1-3: The overview maps (a) are difficult to interpret. Add contour lines or colour by elevation. Maybe highlight the location of critical infrastructure below the monitored sites.

Figure 5: The labels are difficult to read. Please increase the font size. Add a scale bar and a north arrow (if the map is rotated). Are the white spots in (c) snow patches or data gaps?

Figure 6: Change the unit to cm or mm per day. A scale bar and north arrow are missing. It is difficult to recognise the mentioned features in (e) and (f).

Technical corrections

None at this stage.

References

Toffanin, P. (2019). OpenDroneMap: The Missing Guide. MasseranoLabs LLC. [https://odmbook.com/]

Groos et al. (2019). The Potential of Low-Cost UAVs and Open-Source Photogrammetry Software for High-Resolution Monitoring of Alpine Glaciers: A Case Study from the Kanderfirn (Swiss Alps), 9(8), 356. https://doi.org/10.3390/geosciences9080356.

Walter et al. (2022). Brief communication: An autonomous UAV for catchment-wide monitoring of a debris flow torrent. NHESS 22, 4011-4018. https://doi.org/10.5194/nhess-22-4011-2022.