the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Regional-scale analysis of weather-related rockfall triggering mechanisms in Norway, and its sensitivity to climate change
Abstract. This paper evaluates the relation between rockfall events and weather conditions for two regions in Norway – Romsdalen and Gudbrandsdalen and explores how rockfall frequency might change with future climate conditions. Our analysis focuses on understanding the relationship between rockfall occurrence and effective water inputs, including rainfall and snow melt, and temperature oscillations both in cold conditions (freeze-thaw cycles) and in warm conditions (hot-cold cycles). To accomplish this, regional weather data and rockfall information in the Norwegian Mass Movement Database have been employed. Our results indicate that temperature oscillations might be better suited than effective water input to depict the occurrence of rockfalls in the two study areas in Norway. Freeze-thaw cycles are most frequent during winter and spring, and hot-cold cycles are most frequent during summer. Climate change will affect rockfall seasonality and the frequency in which freeze-thaw cycles and hot-cold cycles are observed. Thus, altering the exposure of population and infrastructures to rockfalls.
- Preprint
(2065 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on nhess-2024-46', Anonymous Referee #1, 26 Apr 2024
Review on "Regional-scale analysis of weather-related rockfall triggering mechanisms in Norway, and its sensitivity to climate change" by
Rosa M. Palau, Kjersti Gleditsch Gisnås, Anders Solheim, and Graham Lewis GilbertGeneral comments:
The authors aim at identifying the weather pre-conditions that promote rockfall in two Norwegian regions. The study is within the scope of NHESS. The figures are of good quality and the language is clear.
However, there is a fundamental problem with the method used in the analysis. The relationship between the hot-cold/freeze-thaw cycles and the rockfall events, the authors believe to find, is just a result of the analysis set up. The same holds for the analysis of the relationship between effective water input accumulation and rockfall events.
The method includes a pre-selection of days that enter the analysis. This introduces a bias on the results. Basically, a number of days with certain weather pre-conditions are selected (e.g. 2 temperature cycles in 3 days). Then the pre-conditions for these days are compared to the pre-conditions of all other days. Of course the probability to find the pre-conditions is higher in the selection (100%), compared to the other days. It is completely irrelevant if the selected days are also associated with rockfall events. One would get the same result if a number of non-rockfall days with these pre-conditions would be used as the reference against which all other days are compared.
That the number of temperature cycles increases, when the pre-condition period is extended, is also a result of the analysis design (it cannot decrease).
A fair test if the distribution for these pre-conditions differs between rockfall and non-rockfall days would be the following boot-strap test:
n-pre-ev=number of days with rockfall event and selected pre-conditions
n-pre=number of days with selected pre-conditions
Randomly draw n-pre=n-pre-ev days with the same preconditions but without an event. Repeat this very often (e.g. 1000 times). Compare the distribution from the random days with the distribution for the event days. It there a statistically significant difference (e.g. median of distribution of pre-conditioned rockfall-event days is outside 95% of distribution obtained with the pre-conditioned random days)?
Depending on the results of a statistically sound new analysis, sections 6 and 7 need to be redone.
I am not sure if the required work can be finished within the mayor-revision time frame. It might be necessary to withdraw the paper and resubmit it again later.Major Comments:
- L172 a) Generally means before all events? b) Is this really a special condition before rockfall events or just climatology. How often do these conditions occur when no rockfall event follows?
- L174 Please give the exact definition of how you define a hot-cold cycle. What is a significant temperature oscillation?
- L205 Before (L172) you identified one cycle in the previous 7 days. Why do you now choose two cycles in 3 days as a threshold?
- Figure 6 and 7: By definition there should not be any events in pannels a) and c) with numbers <2 and a period length of 3 days. The boxes, however, indicate that such events exist in the data subset. Why?
- L237 Is the daily temperature cycle in Gudbrandsdalen more pronounced than in Romsdalen? This can be expected because of the larger distance from the ocean.
- L281 "and at least two freeze-thaw cycles have been observed" This is not a result but just reflects the initial event-selection threshold.
- L309 The temperature increase depends on the analysed climate projection. Which greenhouse gas concentration pathway 1.6°C corresponds to?
- L314 Please explain "recommended climate factors"
- L333 Please give more information on the climate simulations and the downscaling. The reviewer was not able to access Hanssen-Bauer et al., 2015 (too little information in references to find it)
Minor comments:
- L 495 Where is this published? Please provide more information
Citation: https://doi.org/10.5194/nhess-2024-46-RC1 -
RC2: 'Comment on nhess-2024-46 - Major revisions needed', Anonymous Referee #2, 22 May 2024
General comment
The work aims to study the relationships between rockfall occurrences and different antecedent weather conditions in two study areas in Norway. The study is within the scope of NHESS and can be interesting for a wide audience. The language is clear and the structure is good. However, I found some issues in the data used, in some parts of the methods and in particular in the discussion related to the evaluation of the influence of projected climate change on these phenomena.
First, I think that using daily rainfall data can significantly hamper the analysis and the results. I would recommend using data with a finer temporal resolution.
Second, I think that some analyses performed by the authors introduce subjectivity and biases in the results, and in some cases drive the results to some direction that is pre-fixed by the used methods.
Third, the selection of “non-events” is quite problematic and it should be fixed in order to obtain reliable results.
Fourth, the whole evaluation of the influence of climate change on rockfall occurrence is made only in a qualitative way. Quantitative evaluations are needed.Then, I found some other specific issues, which I list below.
For all these reasons, I would recommend that the manuscript be substantially revised before it is reconsidered for publication.
Specific comments
The authors refer in several parts of the manuscript to some unspecified “rockfall events”. It is not clear what such rockfall events are. A clarification is needed.
There is some confusion in some parts of the text between “weather” and “climate”. I would suggest to check the text and use the two terms correctly.
The fact that freeze-thaw cycles are most frequent during winter and spring, and hot-cold cycles are most frequent during summer, which is reported by the authors as a result of the analysis, seems simply linked to the climate and weather conditions of the study areas.
The combination between rainfall and snowmelt is not very well described. I would suggest adding some additional details.
I don’t understand why the authors used daily rainfall data. This temporal resolution is coarse, particularly if compared with the 3-h resolution of the temperature. Some explanations are needed.
Regarding the rockfall dataset used, it is not clear what temporal details are included. Is only the day of occurrence of the rockfalls known?
Moreover, the authors say that rockfalls happening in remote areas are unreported. Is it really so strict, or they are just under-represented in the dataset?The authors decided to move the points representing the rockfalls from the deposition to the release area. This is reasonable; however, I think that this introduced some uncertainty and subjectivity in the analysis. This should be acknowledged. Moreover, I would like to read in the manuscript, what’s the mean (minimum, maximum, …) distance between deposition and release locations. And also, the mean (minimum, maximum, …) elevation difference. These figures could be used to give a partial estimation of the uncertainty related to this action.
In lines 155 onward, the authors analyze the differences in the elevation of rockfalls. First, this is strongly influenced by the action of moving the rockfall points from deposition to release zones. Secondly, this should be also related to possible differences in the elevations in the two areas. So, I would suggest elaborating a bit about this issue.
In lines 165 onward, the authors look for some weather-related triggering mechanisms. It is not very clear to me what is the time period over which the time series were analyzed and what the authors mean with “visually examined”. I’m afraid this can introduce other uncertainties and subjectivity in the analysis.
Indeed, then the authors write that freeze-thaw cycles are the most frequent triggering factor (line 177). In this case too, it is not clear how this finding was obtained. Seems again a bit subjective – quantitative measures would be needed.In section 5, a new analysis is presented, which seems different from (or at least not linked to) the analysis reported in section 4.1. It is not very clear how the freeze-thaw cycles were identified and how this new analysis refers to the previous one.
The definition and selection of the “non-event” is quite tricky. I mean, the authors acknowledge that the dataset is not complete. So selecting all the days for which they do not have information about rockfall occurrence could result in a huge overestimation of the non-event days. In a regional- or national-scale dataset, like the one used in this work, the lack of information on rockfall occurrence, does not mean that the rockfall have not occurred. This issue should be better addressed by the authors. Moreover, there are several examples in the literature of methods for defining and selecting “non-event” cases and evaluating the uncertainties resulted from under/over sampling (e.g. https://doi.org/10.5194/nhess-23-2737-2023, https://doi.org/10.5194/nhess-23-1483-2023, https://doi.org/10.1016/j.enggeo.2018.07.019, https://doi.org/10.1016/j.geomorph.2014.10.019).
In lines 295-296 (and looking at figure 10) I do not see the differences in effective water input between the two zones, as written by the authors. Actually, I see only minor differences in terms of effective water input. On the other hand, I see more marked differences for freeze-thaw and hot-cold cycles.
In section 6, the evaluation of the role of future climate scenarios on rockfall occurrence and distribution is made only in a qualitative way. I recommend to search for some relationship (e.g. empirical relationships, correlations, regressions, …) among climate variables and triggering variables (or rockfall occurrence) in the present, in order to use them together with climate projections to look for quantitative projections of changes in rockfall occurrence/distributions (see e.g. https://doi.org/10.1016/j.scitotenv.2017.03.103). Moreover, the use of only one climate scenario (in particular if only the extreme RCP8.5 scenario is selected) is not advisable in such types of analysis, in order to reduce the uncertainties linked to climate projections and obtain more reliable results. Finally, there is an interesting discussion about the RCP 8.5 and how it can’t be longer considered “business-as-usual”: https://www.nature.com/articles/d41586-020-00177-3.
Only in the discussion, in line 396, the authors acknowledge that they couldn’t identify any of the weather-related factors for 50% and 32% of the rockfalls in the two study areas. This means that all the analyses reported in the previous section were made with much less records than the ones reported in the Data section. Something that should not be neglected. Moreover, it is not clear what the authors mean with “biological” activity, in the line below.
Figures
Figure 1: I would suggest adding the coordinate frames. Moreover, please correct the scalebar to common “round” values, e.g. 50, 100, …500 km.
Figure 2: perhaps two tendency curves (e.g. moving average) could be added, for increasing the readability of the two different trends.
Figure 3: why density instead of percentages? Please check the font size.
Figure 4: the meaning of the dotted black lines is not specified.
Figure 5: perhaps this figure could be reshaped to fit in a column (too much empty space now). Moreover, please check the font size. Finally, in the caption I read “Identified weather-related triggering mechanism”. Perhaps “Attributed” would be more appropriate?
Figures 6, 7, 8: the “green triangles” mentioned in the caption are actually diamonds. I would add two labels on top indicating “rockfall presence” and “rockfall absence” panels (or events, non-events, like in figure 9 and 10).
Figures 9, 10: it seems that the darkest color in the graph are darker than the maximum value in the legend on the right. Anyway, I would add the maximum value in the scale/legend.
Figures 11, 12: please replace “10 climate scenarios” with “10 climate projections”, given that all the projections are made under the RCP8.5 scenario, as mentioned by the authors in the text.
Technical corrections
Line 9: I would avoid the repetition of “conditions”
Line 35: I would write “at the regional scale”
Lines 60-63: these sentences should be moved to the discussion or conclusions sections.
Line 133: please correct “provable”
Line 268: please correct “section 0”
Line 279: please correct “Figure (Figure 9 a and b)”
Line 330: I would remove “are” at the end of the first sentence.
Line 390: repeated reference
Citation: https://doi.org/10.5194/nhess-2024-46-RC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
319 | 87 | 26 | 432 | 20 | 20 |
- HTML: 319
- PDF: 87
- XML: 26
- Total: 432
- BibTeX: 20
- EndNote: 20
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1