the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Detecting and discriminating rockfall signals from anthropogenic noise using Distributed Acoustic Sensing along a Norwegian railway
Abstract. In mountainous regions, railway lines frequently traverse terrain with elevated rockfall and avalanche risk. In recent years, several such events have struck railways, resulting in casualties and significant infrastructure damage. Continuous monitoring along railway corridors is therefore essential for early warning and preventive measures. Given that fibre-optic telecommunications cables are already installed along many railways, Distributed Acoustic Sensing (DAS) represents an operationally attractive solution for continuous, large-scale rockfall monitoring. Avalanches and rockfalls release seismic energy as they propagate, which can be recorded by seismic sensors installed along the railway. Here, we use a fibre-optic cable running parallel to a railway in Norway to acquire DAS data and characterise the seismic signatures of rockfall events. The DAS system also recorded a range of anthropogenic signals, including those from road vehicles and trains, and the frequency distribution are found to be distinctly different from rockfalls. Rockfall signals are characterised by higher dominant frequencies and apparent propagation velocities than anthropogenic noise sources. Furthermore, moving vehicles generate persistent low-frequency energy that is absent in rockfall signals, providing an additional basis for discrimination. In addition to natural rockfall events, we carried out controlled rockfall experiments in which rocks of various sizes were dropped along the railway. The smallest rockfall detected had a mass of approximately 50 kg and was observable over approximately 400 m of fibre, whereas the largest boulder could be observed over several kilometres. Combined, these characteristics indicate that rockfalls can be effectively discriminated from anthropogenic signals, forming a promising foundation for the development of automated detection algorithms using cost-efficient DAS system.
- Preprint
(60987 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2026-4460', Anonymous Referee #1, 01 Sep 2026
-
AC1: 'Reply on RC1', Robin André Rørstadbotnen, 28 Oct 2026
General response: We thank the reviewer for the careful reading of our manuscript and for the detailed and constructive comments. We agree that the section "Methods and signal classification" lacks important details. In the revised manuscript we will expand and restructure this section so that it is clear which processing steps, with which parameters, are applied to which data and event types. We will also revise the figure captions throughout so that each figure can be read on its own. We believe these changes will make the manuscript considerably more reproducible and easier to follow. Our responses to the individual comments are given below.
Comment 1: We will add more details on the acquisition system as suggested. As is generally the case for OptoDAS, the noise floor increases with distance from the interrogator, especially after 70-80 km. We will add a supplementary figure showing the noise level along the full 98 km together with some of the observed events to illustrate their signal-to-noise ratios. Based on this we will be able to state the detectability of the different rockfalls at different locations along the fibre.
Comment 2: The tap tests were mainly used to map DAS channels to positions along the track. The higher-risk sections were identified from the knowledge of the Bane NOR driver, Bane NOR’s hazard assessments based on historical event records and official avalanche risk databases. We will rephrase this to make this clear.Comment 3: We distinguish fibre coupling and fibre orientation based on the location of the various events. All rockfalls were <20 m from the railroad and the fibre orientation will only have an effect on near-offsets, we will clarify this. As for trains and cars they are mainly parallel to the fibre and will therefore not be affected by changes in fibre orientation. The only exception is when the European road diverges from the fibre/railroad. We will clarify this in the paper. Finally, earthquakes will be affected by this fibre orientation. However, we are looking at the sediment resonance, which is surprisingly independent of the distance to event, magnitude and azimuthal direction. We will add supplementary figures that shows this and add a relevant reference.
Comment 4: We will expand the "Data processing" subsection and list the relevant steps in the order they are applied, including which anti-aliasing filter was used, the filter phase, type and order.The 5–70 Hz (rockfall) and 0.15–2 Hz (traffic) bands were chosen based on the spectral content observed in the data. We will clearly state how these were chosen.
Comment 5: We agree that the different frequency bands used for data processing and hyperbolic fitting need a clearer explanation. Most of the rockfall energy is between 5 and 50 Hz. Increasing the upper limit from 50 to 70 Hz has little effect on the results. For consistency, we will use 5-70 Hz for both data processing and hyperbolic fitting in the revised manuscript and describe the parameter choices more clearly
Comment 6: We agree. We will reorganise the "Methods and signal classification" section to clearer state which processing steps were used for which event type.
Comment 7: This is a good point to discuss. The fibre is installed in cable duct next to the railroad for the entire 98 km of fibre, so there should be little variation in fibre coupling and burial depth along the 98 km studied. As for the fibre orientation, this does not change as the train moves along the rail as the fibre is always running parallel to it at a distance of 3 to 4 m. However, as already mentioned in comment 3, earthquakes will be affected by the changing orientation but seems to give a repeatable response independent of event magnitude, azimuth direction and magnitude. We will add a panel to the supplementary figure suggested in comment 3 focusing on tunnels. We will add a clearer discussion on these fibre conditions.Comment 8: We will describe how the predicted curves in Fig. 7b and 7d were computed, including the equation and how the amplitude scaling factor A0 was determined.
Comment 9: We agree, a natural rockfall is a more complex source than a single controlled drop. This together with spatial variation along the fibre makes it hard to identify the various seismic phases in the f-k domain. However, they are clearer in the t-x domain. We will add a supplementary figure showing the different seismic phases and why/what we identify them to be (using mainly their apparent velocities). Moreover, we use open access data bases (course, but it gives insightful information) to map the areas with lower strain amplitudes. In the manuscript, we will clearer state how this source together with evidence from the DAS data support near-surface geology as the dominant cause.
Follow this link (https://www.ngu.no/en/geologiske-kart) to see the open access map and select ‘surficial deposits’ and zoom to Fauske, Rognan, Røkland or Lønsdalen.
Comment 10: We will describe how the automatic fits and manual overlays were obtained. The automatic fit minimizes the differences between picked and modelled arrival times using Huber loss, which reduces the influence of outliers. The manual overlays illustrate coherent arrivals visible in the data and give visual estimates of apparent velocity, rather than independent reference measurements. We will assess uncertainty in the automatic estimates by varying the picking settings and reporting how the fitted parameters change.
For the controlled single drop, the automatic fit gives 1240 m/s, whereas the manual overlay of an earlier coherent arrival gives 2200 m/s. A weak first arrival followed by stronger later arrivals can make the automatic picks follow different arrivals on different channels, which makes the fitting off. The low MAE indicates a close fit to those picks but does not confirm that they follow the intended arrival. We will explain why the manual overlay was used and make this limitation clear in the discussion.
Comment 11: There where some miscommunication during the field and we later figured out that we did not use a Robel, but a Windhoff. LTR17 of 14.24 m length. We will update the manuscript.
Comment 12: We will state which unit is used in Fig. 2.
Comment 13: We will revise Fig. 10 so that all panels use a consistent representation.
Comment 14: We will revise all figure captions so that each figure can be interpreted independently, stating the plotted quantity, units, normalisation, frequency band and event shown.
Citation: https://doi.org/10.5194/egusphere-2026-4460-AC1
-
AC1: 'Reply on RC1', Robin André Rørstadbotnen, 28 Oct 2026
-
RC2: 'Comment on egusphere-2026-4460', Anonymous Referee #2, 21 Sep 2026
Recommendation: Reject
The manuscript does not demonstrate its central claim: reliable discrimination of rockfall signals from anthropogenic noise. It presents selected recordings from two natural events and two controlled experiments, together with selected traffic examples, but provides neither a validated discrimination procedure nor sufficient experimental information to reproduce the analysis. More seriously, its principal quantitative comparison mixes vehicle translation speeds with seismic-wave propagation speeds. The processing also depends on the event class already being known.
The topic is relevant to NHESS. However, the manuscript does not establish a transferable result for hazard monitoring or provide the acquisition documentation and methodological validation needed by the DAS community. The concerns below affect the scientific argument itself. Addressing them would require substantial reanalysis and validation, rather than clarification and editing alone. I therefore recommend rejection.
-
The central velocity comparison measures different physical quantities (Section 3.2; lines 259–274; Table 1).
The automatic fits return 24–25 m/s for cars and trucks, corresponding to vehicle movement, and 1240–2642 m/s for rockfalls, interpreted as seismic propagation velocities. The manuscript explicitly acknowledges this distinction, yet subsequently presents the approximately two-order-of-magnitude separation as evidence supporting discrimination. This is not a comparison of seismic-wave velocities between source classes.
Using the manually assigned traffic-wave velocity of 430 m/s instead gives ratios of approximately 2.9–6.1 relative to the automatic rockfall estimates. Those comparisons still require consistent phase identification and control of site effects. Source speed and wave speed might both be useful features, but they must be defined, extracted, and evaluated separately.
The controlled single drop also yields 1240 m/s automatically against 2200 m/s by manual overlay, despite a normalized MAE of 0.01. The manual value is not independent ground truth, but this large disagreement demonstrates that a small fitting residual does not establish a correct wavefront or velocity. The conclusion that rockfall velocities exceed 2 km/s then contradicts the automatic result in Table 1.
-
The fitting procedure uses the event label before the claimed discrimination and lacks uncertainty analysis (lines 105–130).
The fractional-change operation is applied specifically to rockfalls, and their fits receive onset-time and source-offset constraints derived from prior knowledge. Vehicle signals are treated differently. How would an unknown event be processed without first knowing its class? A workflow conditioned on known labels cannot establish classification performance unless a procedure for selecting among those alternatives is defined and independently tested.
The assertion that independent channel picking makes the method insensitive to coupling and geology is also unjustified. These factors affect waveform shape and signal-to-noise ratio, not merely an overall amplitude scale. Huber loss cannot correct systematic selection of later arrivals or mixed phases.
Furthermore, Equation (1) assumes an appropriate source–receiver geometry and an effective constant velocity. Its applicability to curved fibre and heterogeneous terrain must be assessed. Report fitted positions and offsets, parameter uncertainties, and errors against the known controlled-source locations. Normalizing MAE by the analysis-window duration does not supply this information and makes scores from different windows difficult to compare.
-
Reliable discrimination is asserted without being tested, while the most relevant counterexamples are deferred (lines 58–64 and 275–309).
The manuscript gives no explicit decision rule, independent test set, false-alarm rate over representative continuous data, or assessment of missed events. A long deployment does not compensate for a small, selectively presented event sample. State how much data were examined, how events were selected, and how their labels were independently established. Photographs of deposited rocks do not by themselves establish the timing and identity of the selected DAS signals.
Most importantly, lines 294–300 acknowledge unexplained impulsive signals with rockfall-like hyperbolic moveout, but leave their characterization to future work. These are precisely the negative examples needed to evaluate the central claim. Comparing rockfalls mainly with prolonged vehicle passages avoids the difficult discrimination problem.
Nor does the presence of multiple impacts establish a unique rockfall signature. Repeated anthropogenic impacts must be considered. Restricting detection to mapped steep slopes cannot replace signal validation: anthropogenic events also occur near slopes, and spatial exclusion can hide missed hazards. Include these confounding signals in the analysis and quantify the resulting errors.
-
Acquisition, source characterization, and processing are insufficiently documented (Sections 2.1, 2.2, and 3.1).
Naming OptoDAS is not a reproducible acquisition description. Provide the manufacturer and configuration, channel spacing, relevant instrument response, and an event-by-event record of gauge length and sampling rate. Explain when and why 8 or 16 m gauges and 400, 800, or 1000 Hz sampling were used. Section 3.1 describes only resampling from 800 to 200 Hz, leaving the other acquisition settings unresolved.
The fibre installation is equally important: cable construction, burial or duct/tray placement, rail and road offsets, tunnel mounting, slack loops, and coupling conditions are inadequately described. These omissions directly affect the amplitude, velocity, and geological interpretations.
“Sledgehammer shots” sounds more like refreshments on the maintenance journey than a reproducible source description. Specify whether these were manual hammer blows or mechanically generated impacts, and document the equipment, impact surface, source position, stop spacing, and number of strikes. A stopping vehicle is not an impulsive tap; explain its separate role in channel mapping. The quoted ±50 m and ±25 m uncertainties also require a coordinate-survey method and an actual error assessment.
Finally, provide the phase-to-strain conversion, integration treatment, anti-alias filtering, filter order and phase response, tapering, and f–k normalization. These details are necessary to evaluate the reported results.
-
The spectral comparisons do not separate source characteristics from measurement and processing effects (Figures 5–11).
Changing gauge length changes the spatial response and can affect slow and fast apparent waves differently at the same frequency. This is directly relevant when traffic and rockfall spectra are compared. Without assigning acquisition settings to each event and accounting for their response, the observed contrast cannot simply be attributed to source type.
Figure 11 also uses different colour scales and visibly different low-frequency resolutions. Report the time windows, spatial apertures, spectral normalization, decibel reference, and effective resolution. A short impact window cannot be compared with a long vehicle passage without addressing those differences. Explain whether spatial-median subtraction preserves the low-frequency, low-wavenumber physical signals used in the interpretation.
The manuscript further conflates absence of a moving-source ridge with absence of low-frequency energy. These are different claims; low-frequency power is visible in the rockfall panels. The reported velocity ranges also overlap: vehicle-associated energy extends to 2.2 km/s, while natural rockfall energy extends down to 0.5 km/s. Quantify distributions and overlap rather than infer distinct classes from selected images.
-
Geological interpretation and hazard screening are unsupported, and the NGU citation is inadequate (lines 83–95, 133–150, and 316–320).
The cited NGU map link does not identify the actual evidence used for the manuscript’s claims. It provides neither a reproducible account of the hazard delineation nor sufficient support for local deposit thickness and fibre–ground contact. Identify the exact layers or map extracts, scale, version, relevant units, and surveyed locations. Show how these data support each inference. For hazard screening, provide explicit slope/runout criteria and an independent inventory. For cable coupling, provide installation evidence. A generic map link cannot substitute for either.
The promised tap-test characterization of sensitivity versus geology is not presented. Qualitative tunnel/open-section amplitude differences do not fulfil that objective. A tunnel cable is not necessarily directly coupled to bedrock, and lower recorded amplitude does not demonstrate the absence of surface waves or trapped energy. Installation, orientation, coupling, and geology remain confounded.
The conclusion nevertheless attributes a clear geological correlation to the tap-test. That claim requires results which the manuscript does not show.
-
The physical interpretation of directional sensitivity and wave types is not established (lines 247–258).
The statement that incidence near 90° produces maximum axial-strain sensitivity requires an explicit angle definition and derivation. If measured relative to the fibre axis, this is inconsistent with the usual longitudinal plane-wave response, which varies as cos²θ. Rayleigh-wave sensitivity additionally depends on polarization and propagation direction. Source offset alone does not establish the claimed amplitude relationship.
Likewise, apparent velocities of approximately 1.2–2.6 km/s do not, by themselves, identify body waves. Short duration and small rock size do not demonstrate negligible surface-wave generation. Support the phase assignments with independent observations, an appropriate velocity model, or forward modelling.
The claim that rockfalls do not move along the fibre is also an unjustified generalization. Impact positions evolve downslope, with a geometry-dependent projection onto fibre chainage. Conversely, passing vehicles can generate localized impulses at fixed features. These possibilities must be tested before stationary versus moving signatures are treated as reliable class distinctions.
-
Detectability and warning claims exceed the demonstrated evidence (lines 285–309 and conclusions).
The smallest observed rock is not a minimum detectable mass. Its recorded footprint depends on impact energy, substrate, offset, coupling, processing, and background noise. Define a reproducible detection criterion before comparing the reported extents and durations. Kilometres of responding fibre are also not equivalent to kilometres of stand-off detection range.
Train-related saturation or masking is a central operational limitation. Establish whether the observed problem is actual instrument saturation, phase-retrieval failure, or signal masking, and quantify its spatial extent, duration, and frequency. The monitoring system’s blind intervals occur when trains are present; this cannot simply be deferred while effective railway monitoring is advertised.
Finally, distinguish detection after material reaches the track from detection before impact or prediction of failure. No demonstrated latency, location accuracy, or actionable warning time supports the present claims. The data and code needed to assess these results should be available during review, not promised only after acceptance.
The manuscript establishes that selected rockfall events were recorded on existing railway fibre. It does not establish reliable discrimination, validated localization, or an operational warning capability. Because the principal quantitative argument is inconsistent and the decisive validation is missing, I recommend rejection.
Citation: https://doi.org/10.5194/egusphere-2026-4460-RC2 -
AC2: 'Reply on RC2', Robin André Rørstadbotnen, 28 Oct 2026
General answer: We thank the reviewer for the detailed comments. The comments have helped us identify where the wording of the manuscript goes beyond what the results of the paper show.
The manuscript was intended as a characterisation study. Its aim is to document how rockfall signals recorded on a railway fibre differ from the anthropogenic and natural signals recorded on the same fibre, as a basis for future detection work. The reviewer concludes that the manuscript establishes that selected rockfall events were recorded on existing railway fibre. It does more than that: it documents controlled rock drops at known times and locations and two natural rockfall events, together with the traffic, train and earthquake signals recorded on the same fibre, and analyses how these signals differ. Its main quantitative result, the difference in apparent seismic velocity between the surface waves generated by vehicles and the rockfall arrivals, holds once the travel speed of the vehicles is separated from the seismic velocity (Comment 1).
We recognise, however, that the title, the abstract and parts of the discussion and conclusions suggest discrimination capabilities, which the manuscript does not establish. We agree with the reviewer on this point, and in the revised manuscript we will revise the wording to make it clearer that this is a characterisation study. Specifically, we will:
- change the title to "Characterising rockfall and anthropogenic signals using Distributed Acoustic Sensing along a Norwegian railway"
- revise the abstract, introduction and conclusions so that the observed differences are presented as potential features for future discrimination, and remove unsupported statements
- remove statements that can be confused with claims about early warning and operational monitoring capability, and limit the discussion of an automated system to outlook for future work
- expand the documentation of acquisition, installation, and processing (see also comments from reviewer 1)
Since the paper does not aim to provide a detection or classification system, some of the reviewer's requests fall outside the scope of the paper. Our intention was to characterize signals rather than including principled decision rules for testing on an independent dataset as the reviewer suggests, with false-alarm and missed-event rates, and an assessment of detection latency and warning time. We do, however, agree that the manuscript must limit scope more clearly and state these limitations and paths for future work.
The reviewer argues that addressing the concerns "would require substantial reanalysis and validation, rather than clarification and editing alone". The validation requested concerns an automatic detection system, which this study does not claim to provide. The remaining concerns can be addressed through clarification, additional documentation and limited further analysis of the existing data, as described in our responses below. We therefore believe that, with the revisions outlined here, the manuscript merits revision rather than rejection and remains a useful contribution to NHESS.
Comment 1: We agree and thank the reviewer for pointing this out. The fitted velocities for the car and truck (24–25 m/s) describe the speed of the vehicles, not the velocity of a seismic wave (although the vehicles also generate seismic waves), so the statement in lines 267–268 that the fitted velocities of vehicles and rockfalls differ by two orders of magnitude compares two different physical quantities. Although the quantities differ, both carry information that can separate vehicles from rockfalls. The dominant signals from vehicle signal reflect a source travelling along the fibre 15-30 m/s, whereas that of a rockfall reflects seismic waves propagating from a localised impact at 1.2–2.6 km/s in our data. As the reviewer also points out, source speed and wave speed may both be useful features, provided they are defined and evaluated separately. We will rewrite this part of the manuscript to make this distinction clear and use it throughout the paper.
For the seismic arrivals, the manual overlays in Table 1 give 430 m/s for waves generated by vehicles and 2.0-2.6 km/s for the selected rockfall arrivals. These are visual estimates, not independent reference measurements. Different phases, source geometry and local ground conditions can affect the comparison, and the broader velocity ranges in the spectral plots overlap. These differences characterise the selected recordings.
For the controlled single drop, the automatic fit gives 1240 m/s, whereas the manual overlay of an earlier coherent arrival gives 2200 m/s. A weak first arrival followed by stronger later arrivals can make the automatic picks follow different arrivals on different channels. So the automatic fit, based on the noisy data, is off with the manual fitting. The low MAE indicates a close fit to those picks but does not confirm that they follow the intended arrival (see comment 2)
We therefore used manual overlays for these events to illustrate this fact. The statement in the conclusions that rockfall velocities exceed 2 km/s refers to this fastest coherent arrival, which has an apparent velocity of 2.0–2.6 km/s in all four rockfall events, although slower arrivals are also present. We recognise that the manuscript does not explain this clearly, and that the statement as written is inconsistent with Table 1. In the revised manuscript we will explain when and why manual overlays were used, assess uncertainty by varying the picking settings (see Comment 2) and rewrite the conclusion accordingly.
Comment 2: As explained in our general response, this is a characterisation study. Hyperbolic fitting is a supporting analysis used to illustrate and quantify the observed arrival patterns, alongside the time and distance plots and the spectral analysis. It is not intended as an automatic detection or classification procedure. The events were identified before fitting, and the processing and constraints used available information about them. Fractional change was used to make the rockfall onsets clearer, while vehicle picks used filtered absolute amplitudes. We will clarify this role in the manuscript and document the processing settings for each event.
We agree that picking on each channel can make the method sensitive to coupling and geology. Even though it reduces sensitivity by taking absolute amplitude differences, changes in waveform shape, background noise and the relative strength of different arrivals can change the selected time. Huber loss used in the fitting reduces the influence of outliers but cannot correct picks that follow a later arrival or a mixture of phases. We will remove the claim of insensitivity.
Equation (1) is used as a simple approximation of the arrivals around a localised impact. For moveout over about 500 m on each side of the impact, we consider a straight section of fibre within this range to be a reasonable local approximation. A constant effective velocity summarises propagation over the selected section; it does not imply that the ground is uniform at every channel. The arrivals do not need to follow the model exactly for the fit to describe their overall pattern.
For the rockfall examples, which use the same picking procedure, normalised MAE remains useful for comparing fit quality relative to the selected event windows. A lower MAE indicates that the selected picks follow the fitted hyperbola more closely. A higher MAE may reflect other arrival patterns, noise or poor coupling that affects the picks, making a single hyperbola less representative. This describes fit quality but does not establish that the selected arrival or fitted velocity is correct. We will also report the fitted origin time, position along the fibre, source offset and velocity. We will assess uncertainty by varying the picking settings and reporting how the fitted parameters change.
Comment 3: We acknowledge that the wording of the manuscript, including the title and parts of the abstract, introduction (lines 62–63) and discussion, was ambiguous and could be read as describing an automatic detection study. As mentioned above, the manuscript is a characterisation study. Its aim is to document the signatures of controlled and natural rockfalls recorded on a railway fibre, together with the anthropogenic and natural signals recorded on the same fibre, and to identify the features that differ between them. A decision rule, an independent test set, false-alarm and missed-event rates over continuous data is a natural next step after this study. The current paper shows alignment of geophysical DAS signals and real-world events that occurred along the fibre, and it lays the groundwork for further studies on decision support tools. We will revise the wording so that the scope is unambiguous.
We will describe the event selection clearly. The controlled experiments were carried out by the authors and BaneNOR personnel at known times and locations. The natural events were identified by BaneNOR personnel and through track inspection, and the DAS signals were associated with them using the provided time and location information. We will state this more clearly.
The impulsive signals with rockfall-like moveout were reported in the manuscript (lines 294–300) because they show where the features documented here are not sufficient. We agree that they are relevant, and in the revised manuscript we will include supplementary figures that describe them. A full, automatic, characterisation of these signals requires a dedicated analysis of the continuous dataset and is considered outside the scope of this work.
Regarding multi-rock events, our point is that several impacts within a few seconds produce a combined signature that differs from a single impulse. We agree that repeated anthropogenic impacts could produce similar signatures, and we will change our statement accordingly. Restricting detections to mapped slopes was proposed as a possible screening step for a future system, not as validation of the signal characteristics. We will also make this clearer.
Comment 4: We agree and will expand Sections 2 and 3.1 to give the necessary information requested by both reviewer. The revised manuscript will include:
- Interrogator and settings. The interrogator and its configuration, including channel spacing, plus a table giving gauge length and sampling rate for each event analysed.
- Cable installation. Cable construction, placed in a cable, typical offset by 2-3 m to the rail and 15-16 m to the E6 road (Sect. 3.2). Mounting in the tunnels is largely the same as outside, but we will provide additional information about it. We will outline how slack loops/fibre coils where handeled in the preprocessing. We will add a supplementary figure showing and explain how these are identified.
- Tap-test sources. The sledgehammer shots were manual strikes with a 5 kg sledgehammer on the ballast next to the fibre, with 5 strikes per stop and stops spaced approximately 5 km apart. The vehicle was used for a first, coarse mapping of channels to positions. The hammer strikes were used to refine this position at selected stops. We will describe this more clearly.
- Positioning. The coordinates were recorded at each stop using a phone GNSS which reported the associated error (normally around 5 m, inside tunnels more due to limited bandwidth). We will include an error analysis of associated uncertainties.
- The processing steps will be further detailed as suggested by the reviewer.
Comment 5: We agree that the gauge length acts as a wavenumber-dependent filter and that, at a given frequency, it attenuates slow waves more than fast waves. However, the main difference between the vehicle and rockfall signals is their apparent velocity, which the gauge length does not change. The gauge-length response multiplies the f–k spectrum by a factor that depends only on wavenumber. It can therefore reduce the amplitude of energy at a given apparent velocity, but it cannot shift energy to a different velocity or create energy where there is none. The vehicle surface waves at around 200–500 m/s, and the fast rockfall arrivals at 2.0–2.6 km/s are thus not an artefact of the acquisition. The gauge length mainly affects the high-frequency part of the slow vehicle waves, and we will change our statement on dominant frequency accordingly, presenting apparent velocity as the main distinguishing feature.
Thanks for pointing out lacking details related to Figure 11. A common absolute colour scale is not practical: the vehicle signals are about 20 dB stronger than the rockfall signals, so most of the rockfall spectra would hardly be visible. The panels already use the same 40 dB dynamic range, but with an offset depending on the level of each signal. In the revised figure we will keep a common dynamic range across panels and state the dB reference used. We will also report the difference in signal level between the panels, so that both the spectral shapes and the relative signal strengths can be assessed. In addition, we will state the time window, spatial aperture, and the effective resolution of each panel, and address the effect of using different window lengths. Finally, we will describe the common-mode noise removal in more detail. The common-mode noise in these data originates mainly near the interrogator. It is therefore estimated as the median across all channels of the full 98 km fibre at each time sample and subtracted from the data. No additional common-mode suppression is applied to the windows used for the f–k analysis. This step removes only the component common to all channels at a given time. The signals analysed here come from sources within about 20 m of the fibre and are only recorded over up to 5 km of fibre. The signals of interest are not of a big enough spatial scale and do not contribute to the common-mode estimate and therefore not affected by it.
We agree that the absence of a moving-source ridge and the absence of low-frequency energy are different claims. We will rephrase this throughout the text, so that the feature is the absence of a moving-source ridge. The velocity ranges quoted in the manuscript describe the full extent of the energy in each f–k panel, and these overlap, as the reviewer notes. The distinguishing feature, however, is not an absence of overlap but the presence of substantial energy at different apparent velocities. For road traffic, the energy is concentrated between 200 and 500 m/s (lines 178–179). The rockfall events also contain energy below 500 m/s but has most of the energy at higher apparent velocities (Fig. 11c, d). The current text does not state this clearly and this will be addressed.
Comment 6: We agree that the reference to the NGU map needs to be more specific, and we will update the manuscript with the appropriate references. At the scale of the NGU map, deposit boundaries cannot be assumed to be accurate for individual channels along the 98 km fibre, and NGU gives only depth information through well information at some points along the profile. These wells provide depth-to-bedrock measurements. We therefore use the maps to interpret the variation in DAS response along the fibre. We will make this clearer and phrase the corresponding interpretations accordingly. A channel-by-channel geological characterisation would require local surveys along the fibre and is beyond the scope of this study, as stated in lines 147–148.
The avalanche-prone sections were identified qualitatively during the tap-test, by noting where the railway runs along steep slopes, and checked against topographic maps and Bane NOR’s empirical knowledge of where the historical events occurred. We recognize that this is not stated clearly in the paper and will fix that in the revised version and also add appropriate reference(s) to areas with elevated rockfall risk.
We agree that the tap-test results on sensitivity versus geology are not sufficiently presented, and that the statement in the conclusions (lines 316–318) that attribute a clear geological correlation to the tap-test is not supported by the results currently presented in the paper. We will address this in the future version of the paper.
As also noted by Referee 1, installation, orientation, coupling and geology are confounded in the comparison between tunnel and open sections. We will describe the cable installation to a level acceptable to BaneNOR, e.g. how the fibre cable is installed along the track outside and inside tunnels. We will also clearer present why we believe near-surface geology is one contributing factor to the recorded amplitudes and how the coupling and installation parameters affect the results.
Comment 7: We agree that the discussion of directional sensitivity in lines 247–252 is imprecise. In the revised manuscript we will clarify the discussion.
We acknowledge that the apparent velocities of 1.2–2.6 km/s do not by themselves identify body waves, and that surface waves on top of a bedrock medium can travel at similar speeds. However, in the t-x domain data various seismic phases travel along the fibre, with different apparent velocities. This plot should have been included in the initial version of the paper, and we will add it to the new version.
The reviewer is correct that a rockfall source is not stationary, as it migrates downslope. However, when comparing it to vehicle signals, it is important to consider how far the source moves along the fibre. At the rockfall sites analysed here, the slopes are nearly perpendicular to the track, so the material moves mainly towards the fibre rather than along it. For the events analysed here, a rockfall source moves at most a few tens of metres along the fibre. Road vehicles, in contrast, travel parallel to the fibre and pass along several kilometres of it in a single passage. In this setting the two do not produce comparable moving-source signatures. In the revised manuscript we will state this explicitly and address the case when the vehicles are moving close to the fibre for a similar distance as a rock migrating downslope. As described in our response to Comment 3, this study does not develop or evaluate a classifier; the moving-source behaviour is documented as a characteristic of the signals recorded in this setting, not as a validated class distinction.
The road part studied does not cross the railway, so vehicles travelling along the E6 do not pass level crossings. Other localised impulses from road features such as bridge joints and potholes occur during a continuous passage and are therefore accompanied by the vehicle's moving-source signature. There are, however, side roads with level crossings which will be addressed. Such a situation with a car crossing provides a moving source perpendicular to the fibre, but they occur in rather flat terrain.
Comment 8: Several points in this comment concern the performance of an operational detection and warning system: detection thresholds, system availability during train passages, and latency, location accuracy and warning time. As described in our response to Comment 3, this study characterises the recorded signals and does not develop or evaluate such a system. We will revise the wording in the abstract, the outlook (lines 301–309) and the conclusions that suggests otherwise. Within this scope, we respond to the individual points below.
The manuscript reports the smallest rockfall recorded during the acquisition period, not a minimum detectable mass. We agree that the recorded footprint depends on impact energy, substrate, offset, coupling, processing and background noise, and the reported extents and durations are therefore specific to these events. To avoid ambiguity, we will replace "detected" with "recorded" in the abstract.
The manuscript identifies passing trains as a fundamental limitation of the current setup, both in the discussion (lines 284–289) and in the conclusions (lines 331–333). With the revised framing, it makes no claim about operational monitoring performance, so quantifying system availability during train passages is outside its scope.
The recorded events are impacts of material at or near the track, and the manuscript makes no claim about predicting slope failure.
Final comment: We thank the reviewer again for the detailed assessment. The reviewer's final summary concerns discrimination, localisation and an operational warning capability, which, as explained in our general response, are not aims of this study.
Citation: https://doi.org/10.5194/egusphere-2026-4460-AC2
-
Status: closed
-
RC1: 'Comment on egusphere-2026-4460', Anonymous Referee #1, 01 Sep 2026
The manuscript presents an interesting application of Distributed Acoustic Sensing (DAS) to rockfall-induced seismic events and anthropogenic events. The study applied processing and analysing approaches on both experimental and observed rockfall events. In general, I think this study has the potential for publication and is relevant to NHESS. However, several important aspects require further clarification, some key information needs to be added to the manuscript, and some interpretations require further discussion and explanation.
My comments are listed below:
- Given the long distance of fibre (~98km), it would be useful to have additional information on the DAS acquisition. For example, I was wondering about the optical setup, and whether the optical amplification or repeaters were used. It would be also very useful to clarify if the signal quality varies with distance from the interrogator. For example, did the authors check the signal-to-noise ratio or sensitivity along different sections of the fibre, especially near the far end of the 98 km fibre? If this variability exists, could the detectability of rockfalls and other events be affected?
- Lines 83-84: it is unclear to me how the tap tests were used to identify the sections with higher avalanche risk. What observable metrics were derived from the tap test and used for this assessment?
- How were the geological effects from the near surface materials distinguished from the effects from fibre coupling and fibre orientation, etc.? How confident can we be that the main cause of the spatial variation in recorded amplitude is near-surface geological factors?
- The subsection “data processing” is too short. A lot of methodological details are missing from this paragraph. Was the low-pass filter applied before the downsampling (for avoiding aliasing)? Why were the 5-70 Hz and 0.15-2 Hz frequency bands selected for the events of avalanches and traffic? Are these frequency bands based on spectral characteristics observed in the data or previous studies? What is the Butterworth filter order and did you apply the zero-phase filtering? Please better specify the preprocessing procedure and organize the section.
- The hyperbolic-fitting workflow would need a more complete description. Why is the 5-50 Hz band used here but not the 5-70 Hz used in the general processing? It would be helpful to indicate how these processing parameters were selected.
- It is not always clear which processing steps and criteria are applied to which types of events (e.g., controlled cases, natural rockfall, avalanches, or other detected signals). Some of this important information appears at different points throughout the text, which makes the overall processing workflow difficult to follow. I therefore suggest reorganizing the part of “Methods and signal classification”.
- The reduced surface-waves amplitudes observed along the tunnel sections are attributed to the absence of surficial deposits beneath the fibre (around Line 140). However, DAS amplitudes can also be strongly influenced by fibre coupling, depth, orientation, and other installation conditions. Could the authors discuss these alternative factors and provide more explanation?
- It is unclear for me how the predicted curves in Figure 7b and 7d were calculated and how were they compared with the observations quantitatively. For example, how was the amplitude scaling factor (A0) determined for the prediction curves? How was the best-fitting function identified (e.g., least-squares fitting)?
- Lines 232-235: It may require more justification for the link between the unclear seismic phases in the f-k plot of the nature rockfall to the near-surface geology. In particular, natural rockfalls may represent more complex and spatially distributed sources than the controlled experiment through multiple impacts, rolling, etc. In addition, spatial variations in DAS coupling or sensitivity can also influence the f-k representation. Could the authors provide more evidence or explanation to support near-surface geology as the dominant cause?
- The apparent velocities reported in the manuscript appear to be plausible, however, the procedure used to estimate them (and also their associated uncertainties) requires further clarification. In the manuscript, it is mentioned that the hyperbolas were manually overlaid on the data to determine the apparent velocities (~Line 263). However, how was the best-fitting velocity selected? What uncertainty is associated with these choices?
It is also mentioned in the manuscript that the difference between the fitted velocity and the manually estimated velocity for the controlled single rock-drop experiment (Table 1) reveals a limitation of the fitting procedure, despite the very low MAE. Could uncertainty estimates, pick-consistency criteria, or other quality metrics for the inferred velocities be provided to help strengthen the analysis? - Line 88: it may help if the author can provide the actual length of the Robel maintenance vehicle.
- Figure 2: was the normalization or scaling applied when plotting? Please clarify the quantity plotted here.
- Figure 10: it is difficult to interpret the figure in the current form and combination. It is difficult for me to understand why the events in panels (a) and (d) are presented differently from those in panels (b), (c), (e) and (f). Please mention the reason for using different representations, or simply presenting them in a more consistent way. Furthermore, additional information is needed to better understand the figure. Please clarify whether the quantity has been normalized, what is “c” in panels (b,c,e,f) and “d” in panels (a,d), what is represented by the colorbar at the right of (c) and (f), and what is “Value” 1 to 0.
A more detailed caption would improve the readability of the figure. - Several figure captions do not provide sufficient information to interpret the figures independently. I recommend that the authors revise the captions throughout the manuscript.
Citation: https://doi.org/10.5194/egusphere-2026-4460-RC1 -
AC1: 'Reply on RC1', Robin André Rørstadbotnen, 28 Oct 2026
General response: We thank the reviewer for the careful reading of our manuscript and for the detailed and constructive comments. We agree that the section "Methods and signal classification" lacks important details. In the revised manuscript we will expand and restructure this section so that it is clear which processing steps, with which parameters, are applied to which data and event types. We will also revise the figure captions throughout so that each figure can be read on its own. We believe these changes will make the manuscript considerably more reproducible and easier to follow. Our responses to the individual comments are given below.
Comment 1: We will add more details on the acquisition system as suggested. As is generally the case for OptoDAS, the noise floor increases with distance from the interrogator, especially after 70-80 km. We will add a supplementary figure showing the noise level along the full 98 km together with some of the observed events to illustrate their signal-to-noise ratios. Based on this we will be able to state the detectability of the different rockfalls at different locations along the fibre.
Comment 2: The tap tests were mainly used to map DAS channels to positions along the track. The higher-risk sections were identified from the knowledge of the Bane NOR driver, Bane NOR’s hazard assessments based on historical event records and official avalanche risk databases. We will rephrase this to make this clear.Comment 3: We distinguish fibre coupling and fibre orientation based on the location of the various events. All rockfalls were <20 m from the railroad and the fibre orientation will only have an effect on near-offsets, we will clarify this. As for trains and cars they are mainly parallel to the fibre and will therefore not be affected by changes in fibre orientation. The only exception is when the European road diverges from the fibre/railroad. We will clarify this in the paper. Finally, earthquakes will be affected by this fibre orientation. However, we are looking at the sediment resonance, which is surprisingly independent of the distance to event, magnitude and azimuthal direction. We will add supplementary figures that shows this and add a relevant reference.
Comment 4: We will expand the "Data processing" subsection and list the relevant steps in the order they are applied, including which anti-aliasing filter was used, the filter phase, type and order.The 5–70 Hz (rockfall) and 0.15–2 Hz (traffic) bands were chosen based on the spectral content observed in the data. We will clearly state how these were chosen.
Comment 5: We agree that the different frequency bands used for data processing and hyperbolic fitting need a clearer explanation. Most of the rockfall energy is between 5 and 50 Hz. Increasing the upper limit from 50 to 70 Hz has little effect on the results. For consistency, we will use 5-70 Hz for both data processing and hyperbolic fitting in the revised manuscript and describe the parameter choices more clearly
Comment 6: We agree. We will reorganise the "Methods and signal classification" section to clearer state which processing steps were used for which event type.
Comment 7: This is a good point to discuss. The fibre is installed in cable duct next to the railroad for the entire 98 km of fibre, so there should be little variation in fibre coupling and burial depth along the 98 km studied. As for the fibre orientation, this does not change as the train moves along the rail as the fibre is always running parallel to it at a distance of 3 to 4 m. However, as already mentioned in comment 3, earthquakes will be affected by the changing orientation but seems to give a repeatable response independent of event magnitude, azimuth direction and magnitude. We will add a panel to the supplementary figure suggested in comment 3 focusing on tunnels. We will add a clearer discussion on these fibre conditions.Comment 8: We will describe how the predicted curves in Fig. 7b and 7d were computed, including the equation and how the amplitude scaling factor A0 was determined.
Comment 9: We agree, a natural rockfall is a more complex source than a single controlled drop. This together with spatial variation along the fibre makes it hard to identify the various seismic phases in the f-k domain. However, they are clearer in the t-x domain. We will add a supplementary figure showing the different seismic phases and why/what we identify them to be (using mainly their apparent velocities). Moreover, we use open access data bases (course, but it gives insightful information) to map the areas with lower strain amplitudes. In the manuscript, we will clearer state how this source together with evidence from the DAS data support near-surface geology as the dominant cause.
Follow this link (https://www.ngu.no/en/geologiske-kart) to see the open access map and select ‘surficial deposits’ and zoom to Fauske, Rognan, Røkland or Lønsdalen.
Comment 10: We will describe how the automatic fits and manual overlays were obtained. The automatic fit minimizes the differences between picked and modelled arrival times using Huber loss, which reduces the influence of outliers. The manual overlays illustrate coherent arrivals visible in the data and give visual estimates of apparent velocity, rather than independent reference measurements. We will assess uncertainty in the automatic estimates by varying the picking settings and reporting how the fitted parameters change.
For the controlled single drop, the automatic fit gives 1240 m/s, whereas the manual overlay of an earlier coherent arrival gives 2200 m/s. A weak first arrival followed by stronger later arrivals can make the automatic picks follow different arrivals on different channels, which makes the fitting off. The low MAE indicates a close fit to those picks but does not confirm that they follow the intended arrival. We will explain why the manual overlay was used and make this limitation clear in the discussion.
Comment 11: There where some miscommunication during the field and we later figured out that we did not use a Robel, but a Windhoff. LTR17 of 14.24 m length. We will update the manuscript.
Comment 12: We will state which unit is used in Fig. 2.
Comment 13: We will revise Fig. 10 so that all panels use a consistent representation.
Comment 14: We will revise all figure captions so that each figure can be interpreted independently, stating the plotted quantity, units, normalisation, frequency band and event shown.
Citation: https://doi.org/10.5194/egusphere-2026-4460-AC1
-
RC2: 'Comment on egusphere-2026-4460', Anonymous Referee #2, 21 Sep 2026
Recommendation: Reject
The manuscript does not demonstrate its central claim: reliable discrimination of rockfall signals from anthropogenic noise. It presents selected recordings from two natural events and two controlled experiments, together with selected traffic examples, but provides neither a validated discrimination procedure nor sufficient experimental information to reproduce the analysis. More seriously, its principal quantitative comparison mixes vehicle translation speeds with seismic-wave propagation speeds. The processing also depends on the event class already being known.
The topic is relevant to NHESS. However, the manuscript does not establish a transferable result for hazard monitoring or provide the acquisition documentation and methodological validation needed by the DAS community. The concerns below affect the scientific argument itself. Addressing them would require substantial reanalysis and validation, rather than clarification and editing alone. I therefore recommend rejection.
-
The central velocity comparison measures different physical quantities (Section 3.2; lines 259–274; Table 1).
The automatic fits return 24–25 m/s for cars and trucks, corresponding to vehicle movement, and 1240–2642 m/s for rockfalls, interpreted as seismic propagation velocities. The manuscript explicitly acknowledges this distinction, yet subsequently presents the approximately two-order-of-magnitude separation as evidence supporting discrimination. This is not a comparison of seismic-wave velocities between source classes.
Using the manually assigned traffic-wave velocity of 430 m/s instead gives ratios of approximately 2.9–6.1 relative to the automatic rockfall estimates. Those comparisons still require consistent phase identification and control of site effects. Source speed and wave speed might both be useful features, but they must be defined, extracted, and evaluated separately.
The controlled single drop also yields 1240 m/s automatically against 2200 m/s by manual overlay, despite a normalized MAE of 0.01. The manual value is not independent ground truth, but this large disagreement demonstrates that a small fitting residual does not establish a correct wavefront or velocity. The conclusion that rockfall velocities exceed 2 km/s then contradicts the automatic result in Table 1.
-
The fitting procedure uses the event label before the claimed discrimination and lacks uncertainty analysis (lines 105–130).
The fractional-change operation is applied specifically to rockfalls, and their fits receive onset-time and source-offset constraints derived from prior knowledge. Vehicle signals are treated differently. How would an unknown event be processed without first knowing its class? A workflow conditioned on known labels cannot establish classification performance unless a procedure for selecting among those alternatives is defined and independently tested.
The assertion that independent channel picking makes the method insensitive to coupling and geology is also unjustified. These factors affect waveform shape and signal-to-noise ratio, not merely an overall amplitude scale. Huber loss cannot correct systematic selection of later arrivals or mixed phases.
Furthermore, Equation (1) assumes an appropriate source–receiver geometry and an effective constant velocity. Its applicability to curved fibre and heterogeneous terrain must be assessed. Report fitted positions and offsets, parameter uncertainties, and errors against the known controlled-source locations. Normalizing MAE by the analysis-window duration does not supply this information and makes scores from different windows difficult to compare.
-
Reliable discrimination is asserted without being tested, while the most relevant counterexamples are deferred (lines 58–64 and 275–309).
The manuscript gives no explicit decision rule, independent test set, false-alarm rate over representative continuous data, or assessment of missed events. A long deployment does not compensate for a small, selectively presented event sample. State how much data were examined, how events were selected, and how their labels were independently established. Photographs of deposited rocks do not by themselves establish the timing and identity of the selected DAS signals.
Most importantly, lines 294–300 acknowledge unexplained impulsive signals with rockfall-like hyperbolic moveout, but leave their characterization to future work. These are precisely the negative examples needed to evaluate the central claim. Comparing rockfalls mainly with prolonged vehicle passages avoids the difficult discrimination problem.
Nor does the presence of multiple impacts establish a unique rockfall signature. Repeated anthropogenic impacts must be considered. Restricting detection to mapped steep slopes cannot replace signal validation: anthropogenic events also occur near slopes, and spatial exclusion can hide missed hazards. Include these confounding signals in the analysis and quantify the resulting errors.
-
Acquisition, source characterization, and processing are insufficiently documented (Sections 2.1, 2.2, and 3.1).
Naming OptoDAS is not a reproducible acquisition description. Provide the manufacturer and configuration, channel spacing, relevant instrument response, and an event-by-event record of gauge length and sampling rate. Explain when and why 8 or 16 m gauges and 400, 800, or 1000 Hz sampling were used. Section 3.1 describes only resampling from 800 to 200 Hz, leaving the other acquisition settings unresolved.
The fibre installation is equally important: cable construction, burial or duct/tray placement, rail and road offsets, tunnel mounting, slack loops, and coupling conditions are inadequately described. These omissions directly affect the amplitude, velocity, and geological interpretations.
“Sledgehammer shots” sounds more like refreshments on the maintenance journey than a reproducible source description. Specify whether these were manual hammer blows or mechanically generated impacts, and document the equipment, impact surface, source position, stop spacing, and number of strikes. A stopping vehicle is not an impulsive tap; explain its separate role in channel mapping. The quoted ±50 m and ±25 m uncertainties also require a coordinate-survey method and an actual error assessment.
Finally, provide the phase-to-strain conversion, integration treatment, anti-alias filtering, filter order and phase response, tapering, and f–k normalization. These details are necessary to evaluate the reported results.
-
The spectral comparisons do not separate source characteristics from measurement and processing effects (Figures 5–11).
Changing gauge length changes the spatial response and can affect slow and fast apparent waves differently at the same frequency. This is directly relevant when traffic and rockfall spectra are compared. Without assigning acquisition settings to each event and accounting for their response, the observed contrast cannot simply be attributed to source type.
Figure 11 also uses different colour scales and visibly different low-frequency resolutions. Report the time windows, spatial apertures, spectral normalization, decibel reference, and effective resolution. A short impact window cannot be compared with a long vehicle passage without addressing those differences. Explain whether spatial-median subtraction preserves the low-frequency, low-wavenumber physical signals used in the interpretation.
The manuscript further conflates absence of a moving-source ridge with absence of low-frequency energy. These are different claims; low-frequency power is visible in the rockfall panels. The reported velocity ranges also overlap: vehicle-associated energy extends to 2.2 km/s, while natural rockfall energy extends down to 0.5 km/s. Quantify distributions and overlap rather than infer distinct classes from selected images.
-
Geological interpretation and hazard screening are unsupported, and the NGU citation is inadequate (lines 83–95, 133–150, and 316–320).
The cited NGU map link does not identify the actual evidence used for the manuscript’s claims. It provides neither a reproducible account of the hazard delineation nor sufficient support for local deposit thickness and fibre–ground contact. Identify the exact layers or map extracts, scale, version, relevant units, and surveyed locations. Show how these data support each inference. For hazard screening, provide explicit slope/runout criteria and an independent inventory. For cable coupling, provide installation evidence. A generic map link cannot substitute for either.
The promised tap-test characterization of sensitivity versus geology is not presented. Qualitative tunnel/open-section amplitude differences do not fulfil that objective. A tunnel cable is not necessarily directly coupled to bedrock, and lower recorded amplitude does not demonstrate the absence of surface waves or trapped energy. Installation, orientation, coupling, and geology remain confounded.
The conclusion nevertheless attributes a clear geological correlation to the tap-test. That claim requires results which the manuscript does not show.
-
The physical interpretation of directional sensitivity and wave types is not established (lines 247–258).
The statement that incidence near 90° produces maximum axial-strain sensitivity requires an explicit angle definition and derivation. If measured relative to the fibre axis, this is inconsistent with the usual longitudinal plane-wave response, which varies as cos²θ. Rayleigh-wave sensitivity additionally depends on polarization and propagation direction. Source offset alone does not establish the claimed amplitude relationship.
Likewise, apparent velocities of approximately 1.2–2.6 km/s do not, by themselves, identify body waves. Short duration and small rock size do not demonstrate negligible surface-wave generation. Support the phase assignments with independent observations, an appropriate velocity model, or forward modelling.
The claim that rockfalls do not move along the fibre is also an unjustified generalization. Impact positions evolve downslope, with a geometry-dependent projection onto fibre chainage. Conversely, passing vehicles can generate localized impulses at fixed features. These possibilities must be tested before stationary versus moving signatures are treated as reliable class distinctions.
-
Detectability and warning claims exceed the demonstrated evidence (lines 285–309 and conclusions).
The smallest observed rock is not a minimum detectable mass. Its recorded footprint depends on impact energy, substrate, offset, coupling, processing, and background noise. Define a reproducible detection criterion before comparing the reported extents and durations. Kilometres of responding fibre are also not equivalent to kilometres of stand-off detection range.
Train-related saturation or masking is a central operational limitation. Establish whether the observed problem is actual instrument saturation, phase-retrieval failure, or signal masking, and quantify its spatial extent, duration, and frequency. The monitoring system’s blind intervals occur when trains are present; this cannot simply be deferred while effective railway monitoring is advertised.
Finally, distinguish detection after material reaches the track from detection before impact or prediction of failure. No demonstrated latency, location accuracy, or actionable warning time supports the present claims. The data and code needed to assess these results should be available during review, not promised only after acceptance.
The manuscript establishes that selected rockfall events were recorded on existing railway fibre. It does not establish reliable discrimination, validated localization, or an operational warning capability. Because the principal quantitative argument is inconsistent and the decisive validation is missing, I recommend rejection.
Citation: https://doi.org/10.5194/egusphere-2026-4460-RC2 -
AC2: 'Reply on RC2', Robin André Rørstadbotnen, 28 Oct 2026
General answer: We thank the reviewer for the detailed comments. The comments have helped us identify where the wording of the manuscript goes beyond what the results of the paper show.
The manuscript was intended as a characterisation study. Its aim is to document how rockfall signals recorded on a railway fibre differ from the anthropogenic and natural signals recorded on the same fibre, as a basis for future detection work. The reviewer concludes that the manuscript establishes that selected rockfall events were recorded on existing railway fibre. It does more than that: it documents controlled rock drops at known times and locations and two natural rockfall events, together with the traffic, train and earthquake signals recorded on the same fibre, and analyses how these signals differ. Its main quantitative result, the difference in apparent seismic velocity between the surface waves generated by vehicles and the rockfall arrivals, holds once the travel speed of the vehicles is separated from the seismic velocity (Comment 1).
We recognise, however, that the title, the abstract and parts of the discussion and conclusions suggest discrimination capabilities, which the manuscript does not establish. We agree with the reviewer on this point, and in the revised manuscript we will revise the wording to make it clearer that this is a characterisation study. Specifically, we will:
- change the title to "Characterising rockfall and anthropogenic signals using Distributed Acoustic Sensing along a Norwegian railway"
- revise the abstract, introduction and conclusions so that the observed differences are presented as potential features for future discrimination, and remove unsupported statements
- remove statements that can be confused with claims about early warning and operational monitoring capability, and limit the discussion of an automated system to outlook for future work
- expand the documentation of acquisition, installation, and processing (see also comments from reviewer 1)
Since the paper does not aim to provide a detection or classification system, some of the reviewer's requests fall outside the scope of the paper. Our intention was to characterize signals rather than including principled decision rules for testing on an independent dataset as the reviewer suggests, with false-alarm and missed-event rates, and an assessment of detection latency and warning time. We do, however, agree that the manuscript must limit scope more clearly and state these limitations and paths for future work.
The reviewer argues that addressing the concerns "would require substantial reanalysis and validation, rather than clarification and editing alone". The validation requested concerns an automatic detection system, which this study does not claim to provide. The remaining concerns can be addressed through clarification, additional documentation and limited further analysis of the existing data, as described in our responses below. We therefore believe that, with the revisions outlined here, the manuscript merits revision rather than rejection and remains a useful contribution to NHESS.
Comment 1: We agree and thank the reviewer for pointing this out. The fitted velocities for the car and truck (24–25 m/s) describe the speed of the vehicles, not the velocity of a seismic wave (although the vehicles also generate seismic waves), so the statement in lines 267–268 that the fitted velocities of vehicles and rockfalls differ by two orders of magnitude compares two different physical quantities. Although the quantities differ, both carry information that can separate vehicles from rockfalls. The dominant signals from vehicle signal reflect a source travelling along the fibre 15-30 m/s, whereas that of a rockfall reflects seismic waves propagating from a localised impact at 1.2–2.6 km/s in our data. As the reviewer also points out, source speed and wave speed may both be useful features, provided they are defined and evaluated separately. We will rewrite this part of the manuscript to make this distinction clear and use it throughout the paper.
For the seismic arrivals, the manual overlays in Table 1 give 430 m/s for waves generated by vehicles and 2.0-2.6 km/s for the selected rockfall arrivals. These are visual estimates, not independent reference measurements. Different phases, source geometry and local ground conditions can affect the comparison, and the broader velocity ranges in the spectral plots overlap. These differences characterise the selected recordings.
For the controlled single drop, the automatic fit gives 1240 m/s, whereas the manual overlay of an earlier coherent arrival gives 2200 m/s. A weak first arrival followed by stronger later arrivals can make the automatic picks follow different arrivals on different channels. So the automatic fit, based on the noisy data, is off with the manual fitting. The low MAE indicates a close fit to those picks but does not confirm that they follow the intended arrival (see comment 2)
We therefore used manual overlays for these events to illustrate this fact. The statement in the conclusions that rockfall velocities exceed 2 km/s refers to this fastest coherent arrival, which has an apparent velocity of 2.0–2.6 km/s in all four rockfall events, although slower arrivals are also present. We recognise that the manuscript does not explain this clearly, and that the statement as written is inconsistent with Table 1. In the revised manuscript we will explain when and why manual overlays were used, assess uncertainty by varying the picking settings (see Comment 2) and rewrite the conclusion accordingly.
Comment 2: As explained in our general response, this is a characterisation study. Hyperbolic fitting is a supporting analysis used to illustrate and quantify the observed arrival patterns, alongside the time and distance plots and the spectral analysis. It is not intended as an automatic detection or classification procedure. The events were identified before fitting, and the processing and constraints used available information about them. Fractional change was used to make the rockfall onsets clearer, while vehicle picks used filtered absolute amplitudes. We will clarify this role in the manuscript and document the processing settings for each event.
We agree that picking on each channel can make the method sensitive to coupling and geology. Even though it reduces sensitivity by taking absolute amplitude differences, changes in waveform shape, background noise and the relative strength of different arrivals can change the selected time. Huber loss used in the fitting reduces the influence of outliers but cannot correct picks that follow a later arrival or a mixture of phases. We will remove the claim of insensitivity.
Equation (1) is used as a simple approximation of the arrivals around a localised impact. For moveout over about 500 m on each side of the impact, we consider a straight section of fibre within this range to be a reasonable local approximation. A constant effective velocity summarises propagation over the selected section; it does not imply that the ground is uniform at every channel. The arrivals do not need to follow the model exactly for the fit to describe their overall pattern.
For the rockfall examples, which use the same picking procedure, normalised MAE remains useful for comparing fit quality relative to the selected event windows. A lower MAE indicates that the selected picks follow the fitted hyperbola more closely. A higher MAE may reflect other arrival patterns, noise or poor coupling that affects the picks, making a single hyperbola less representative. This describes fit quality but does not establish that the selected arrival or fitted velocity is correct. We will also report the fitted origin time, position along the fibre, source offset and velocity. We will assess uncertainty by varying the picking settings and reporting how the fitted parameters change.
Comment 3: We acknowledge that the wording of the manuscript, including the title and parts of the abstract, introduction (lines 62–63) and discussion, was ambiguous and could be read as describing an automatic detection study. As mentioned above, the manuscript is a characterisation study. Its aim is to document the signatures of controlled and natural rockfalls recorded on a railway fibre, together with the anthropogenic and natural signals recorded on the same fibre, and to identify the features that differ between them. A decision rule, an independent test set, false-alarm and missed-event rates over continuous data is a natural next step after this study. The current paper shows alignment of geophysical DAS signals and real-world events that occurred along the fibre, and it lays the groundwork for further studies on decision support tools. We will revise the wording so that the scope is unambiguous.
We will describe the event selection clearly. The controlled experiments were carried out by the authors and BaneNOR personnel at known times and locations. The natural events were identified by BaneNOR personnel and through track inspection, and the DAS signals were associated with them using the provided time and location information. We will state this more clearly.
The impulsive signals with rockfall-like moveout were reported in the manuscript (lines 294–300) because they show where the features documented here are not sufficient. We agree that they are relevant, and in the revised manuscript we will include supplementary figures that describe them. A full, automatic, characterisation of these signals requires a dedicated analysis of the continuous dataset and is considered outside the scope of this work.
Regarding multi-rock events, our point is that several impacts within a few seconds produce a combined signature that differs from a single impulse. We agree that repeated anthropogenic impacts could produce similar signatures, and we will change our statement accordingly. Restricting detections to mapped slopes was proposed as a possible screening step for a future system, not as validation of the signal characteristics. We will also make this clearer.
Comment 4: We agree and will expand Sections 2 and 3.1 to give the necessary information requested by both reviewer. The revised manuscript will include:
- Interrogator and settings. The interrogator and its configuration, including channel spacing, plus a table giving gauge length and sampling rate for each event analysed.
- Cable installation. Cable construction, placed in a cable, typical offset by 2-3 m to the rail and 15-16 m to the E6 road (Sect. 3.2). Mounting in the tunnels is largely the same as outside, but we will provide additional information about it. We will outline how slack loops/fibre coils where handeled in the preprocessing. We will add a supplementary figure showing and explain how these are identified.
- Tap-test sources. The sledgehammer shots were manual strikes with a 5 kg sledgehammer on the ballast next to the fibre, with 5 strikes per stop and stops spaced approximately 5 km apart. The vehicle was used for a first, coarse mapping of channels to positions. The hammer strikes were used to refine this position at selected stops. We will describe this more clearly.
- Positioning. The coordinates were recorded at each stop using a phone GNSS which reported the associated error (normally around 5 m, inside tunnels more due to limited bandwidth). We will include an error analysis of associated uncertainties.
- The processing steps will be further detailed as suggested by the reviewer.
Comment 5: We agree that the gauge length acts as a wavenumber-dependent filter and that, at a given frequency, it attenuates slow waves more than fast waves. However, the main difference between the vehicle and rockfall signals is their apparent velocity, which the gauge length does not change. The gauge-length response multiplies the f–k spectrum by a factor that depends only on wavenumber. It can therefore reduce the amplitude of energy at a given apparent velocity, but it cannot shift energy to a different velocity or create energy where there is none. The vehicle surface waves at around 200–500 m/s, and the fast rockfall arrivals at 2.0–2.6 km/s are thus not an artefact of the acquisition. The gauge length mainly affects the high-frequency part of the slow vehicle waves, and we will change our statement on dominant frequency accordingly, presenting apparent velocity as the main distinguishing feature.
Thanks for pointing out lacking details related to Figure 11. A common absolute colour scale is not practical: the vehicle signals are about 20 dB stronger than the rockfall signals, so most of the rockfall spectra would hardly be visible. The panels already use the same 40 dB dynamic range, but with an offset depending on the level of each signal. In the revised figure we will keep a common dynamic range across panels and state the dB reference used. We will also report the difference in signal level between the panels, so that both the spectral shapes and the relative signal strengths can be assessed. In addition, we will state the time window, spatial aperture, and the effective resolution of each panel, and address the effect of using different window lengths. Finally, we will describe the common-mode noise removal in more detail. The common-mode noise in these data originates mainly near the interrogator. It is therefore estimated as the median across all channels of the full 98 km fibre at each time sample and subtracted from the data. No additional common-mode suppression is applied to the windows used for the f–k analysis. This step removes only the component common to all channels at a given time. The signals analysed here come from sources within about 20 m of the fibre and are only recorded over up to 5 km of fibre. The signals of interest are not of a big enough spatial scale and do not contribute to the common-mode estimate and therefore not affected by it.
We agree that the absence of a moving-source ridge and the absence of low-frequency energy are different claims. We will rephrase this throughout the text, so that the feature is the absence of a moving-source ridge. The velocity ranges quoted in the manuscript describe the full extent of the energy in each f–k panel, and these overlap, as the reviewer notes. The distinguishing feature, however, is not an absence of overlap but the presence of substantial energy at different apparent velocities. For road traffic, the energy is concentrated between 200 and 500 m/s (lines 178–179). The rockfall events also contain energy below 500 m/s but has most of the energy at higher apparent velocities (Fig. 11c, d). The current text does not state this clearly and this will be addressed.
Comment 6: We agree that the reference to the NGU map needs to be more specific, and we will update the manuscript with the appropriate references. At the scale of the NGU map, deposit boundaries cannot be assumed to be accurate for individual channels along the 98 km fibre, and NGU gives only depth information through well information at some points along the profile. These wells provide depth-to-bedrock measurements. We therefore use the maps to interpret the variation in DAS response along the fibre. We will make this clearer and phrase the corresponding interpretations accordingly. A channel-by-channel geological characterisation would require local surveys along the fibre and is beyond the scope of this study, as stated in lines 147–148.
The avalanche-prone sections were identified qualitatively during the tap-test, by noting where the railway runs along steep slopes, and checked against topographic maps and Bane NOR’s empirical knowledge of where the historical events occurred. We recognize that this is not stated clearly in the paper and will fix that in the revised version and also add appropriate reference(s) to areas with elevated rockfall risk.
We agree that the tap-test results on sensitivity versus geology are not sufficiently presented, and that the statement in the conclusions (lines 316–318) that attribute a clear geological correlation to the tap-test is not supported by the results currently presented in the paper. We will address this in the future version of the paper.
As also noted by Referee 1, installation, orientation, coupling and geology are confounded in the comparison between tunnel and open sections. We will describe the cable installation to a level acceptable to BaneNOR, e.g. how the fibre cable is installed along the track outside and inside tunnels. We will also clearer present why we believe near-surface geology is one contributing factor to the recorded amplitudes and how the coupling and installation parameters affect the results.
Comment 7: We agree that the discussion of directional sensitivity in lines 247–252 is imprecise. In the revised manuscript we will clarify the discussion.
We acknowledge that the apparent velocities of 1.2–2.6 km/s do not by themselves identify body waves, and that surface waves on top of a bedrock medium can travel at similar speeds. However, in the t-x domain data various seismic phases travel along the fibre, with different apparent velocities. This plot should have been included in the initial version of the paper, and we will add it to the new version.
The reviewer is correct that a rockfall source is not stationary, as it migrates downslope. However, when comparing it to vehicle signals, it is important to consider how far the source moves along the fibre. At the rockfall sites analysed here, the slopes are nearly perpendicular to the track, so the material moves mainly towards the fibre rather than along it. For the events analysed here, a rockfall source moves at most a few tens of metres along the fibre. Road vehicles, in contrast, travel parallel to the fibre and pass along several kilometres of it in a single passage. In this setting the two do not produce comparable moving-source signatures. In the revised manuscript we will state this explicitly and address the case when the vehicles are moving close to the fibre for a similar distance as a rock migrating downslope. As described in our response to Comment 3, this study does not develop or evaluate a classifier; the moving-source behaviour is documented as a characteristic of the signals recorded in this setting, not as a validated class distinction.
The road part studied does not cross the railway, so vehicles travelling along the E6 do not pass level crossings. Other localised impulses from road features such as bridge joints and potholes occur during a continuous passage and are therefore accompanied by the vehicle's moving-source signature. There are, however, side roads with level crossings which will be addressed. Such a situation with a car crossing provides a moving source perpendicular to the fibre, but they occur in rather flat terrain.
Comment 8: Several points in this comment concern the performance of an operational detection and warning system: detection thresholds, system availability during train passages, and latency, location accuracy and warning time. As described in our response to Comment 3, this study characterises the recorded signals and does not develop or evaluate such a system. We will revise the wording in the abstract, the outlook (lines 301–309) and the conclusions that suggests otherwise. Within this scope, we respond to the individual points below.
The manuscript reports the smallest rockfall recorded during the acquisition period, not a minimum detectable mass. We agree that the recorded footprint depends on impact energy, substrate, offset, coupling, processing and background noise, and the reported extents and durations are therefore specific to these events. To avoid ambiguity, we will replace "detected" with "recorded" in the abstract.
The manuscript identifies passing trains as a fundamental limitation of the current setup, both in the discussion (lines 284–289) and in the conclusions (lines 331–333). With the revised framing, it makes no claim about operational monitoring performance, so quantifying system availability during train passages is outside its scope.
The recorded events are impacts of material at or near the track, and the manuscript makes no claim about predicting slope failure.
Final comment: We thank the reviewer again for the detailed assessment. The reviewer's final summary concerns discrimination, localisation and an operational warning capability, which, as explained in our general response, are not aims of this study.
Citation: https://doi.org/10.5194/egusphere-2026-4460-AC2
-
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 356 | 196 | 111 | 663 | 107 | 104 |
- HTML: 356
- PDF: 196
- XML: 111
- Total: 663
- BibTeX: 107
- EndNote: 104
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript presents an interesting application of Distributed Acoustic Sensing (DAS) to rockfall-induced seismic events and anthropogenic events. The study applied processing and analysing approaches on both experimental and observed rockfall events. In general, I think this study has the potential for publication and is relevant to NHESS. However, several important aspects require further clarification, some key information needs to be added to the manuscript, and some interpretations require further discussion and explanation.
My comments are listed below:
It is also mentioned in the manuscript that the difference between the fitted velocity and the manually estimated velocity for the controlled single rock-drop experiment (Table 1) reveals a limitation of the fitting procedure, despite the very low MAE. Could uncertainty estimates, pick-consistency criteria, or other quality metrics for the inferred velocities be provided to help strengthen the analysis?
A more detailed caption would improve the readability of the figure.