General comments

"Rockfall triggering and meteorological variables in the Dolomites (Italian Eastern Alps)" by Bonometti et al. aims to investigate the relationship between meteorological variables and rockfall occurrence in the Italian Eastern Alps. Specifically, a statistical modelling approach based on previous studies and on the Bayesian method has been exploited to assess the frequency of meteorological variables in the ongoing context of climate change and as potential triggers of rockfall events in the studied area in the period 1970-2019. Alongside, an in-depth analysis of the impact of long-term meteorological trends of air temperature, precipitation, temperature variations and freeze-thaw cycles at different aggregation scales on rockfall occurrence has been provided.

The manuscript represents a valuable and innovative contribution to the understanding of meteorological variables-related impact on the rockfall risk. Results of this study are very interesting and properly compared to previous works on the same topic. Despite not being a mother-tongue, I think that the paper is in general well-written.

However, I believe that some major revisions are needed to enhance the overall quality and clarity of the paper before acceptance for publication.

While the methodology is generally well-explained from a technical point of view, there are some assumptions and flows that need clarification. Further, the paper would benefit from a more logical structure, particularly in the Results and Discussion sections. Some parts of the methodology are discussed in the Results, which is not consistent with the purpose of the section. Similarly, some results are overlooked in the Results section and rather included in the Discussion. The Introduction needs some refinements, to better highlight the aim of the work and its novelty with respect to previous mentioned works.

Further details are given below.

Specific comments

Abstract

L 18: This sentence is not clear, please rephrase.

Introduction

Rather than focusing on limitations of the previous method, it seems that this is addressing a different aim. The mentioned works concentrate on the probability occurrence of extremes in the meteorological variables leading to slope failures, using a non-parametric approach to link meteorological anomalies to landslides occurrence. In contrast, this work shifts the focus to long-term trends and frequency over time, specifically investigating how these trends influence rockfall occurrences, especially considering the impacts of warming and changing weather patterns. I think that the primary aim is to assess how the variation in climate variables over time and space affect changes in rockfall frequency, rather than identifying the role of specific climate extremes as triggering events. I invite the authors to refine this part to better highlight the purpose and novelty of the work.

L62-65 This part is more suited in the methodology section.

Case study

A concise overview of the climate of the study area is suggested. Some text from the discussion should be moved here to better contextualize the study area from a climatological point of view.

L 111 Please provide references for these datasets if available.

L115 Are you thus considering all landslide events in the region regardless of the elevation? Please clarify.

Method

As far as I understand, the method focuses on the frequency of meteorological values within their characteristics value ranges in the period preceding slope failure occurrence. A more detailed explanation of these “characteristics value ranges” and how they are defined should be provided.

The methodology is partially based on Paranunzio et al., particularly the time series sampling approach at different time aggregation scales. Then it differs by applying a Bayesian method to assess the relative influence of a variable to act as a trigger of a rockfall in terms of conditional probability. Thus, it seems that the computation of the non-exceedance probability, using a defined alpha level (as stated in the Introduction in reference to previous works) for the detection of potential anomalous values (in statistical sense) is not fully addressed in this paper. However, the outcomes of this approach are then presented in the discussion (L 553 on, Fig. 21.). This creates some confusion, as the linkage between the methodology and the results is not clearly explained. The paper should provide a clearer explanation of how this method connects to the presented results.

L 127 The concept is not clear to me: meteorological variables like e.g. temperature are continuous variables, thus what does the premise” this method focuses on the frequency of meteorological values” mean exactly?

Eq (4) Define j

L 208 In my opinion, this statement is based on a flawed assumption. The method illustrated in Paranunzio et al. is a statistical approach based on the detection of meteorological anomalies (percentiles). As such, this allows to remove possible bias in the absolute rainfall estimates. Paranunzio et al. compute the probability distribution using the climate data recorded at the reference stations as they are and did not transpose the temperature or precipitation measurements from the meteorological stations to the geographical location and elevation of the rockfall detachment zone. This is because the application of a constant lapse rate (as in the case of temperature) would only shift the values, without influencing the probability estimate associated with V. Therefore, I think that it is not accurate to claim that the method presented in this work addresses an issue overlooked in previous methods. Rather, they provide an alternative approach to handling spatial variability in absolute values instead of percentiles. This point needs to be clarified.

L 221 Generally, an environmental lapse rate which considers air temperature decreasing with height at a rate of approx. 0.6 °C/100 m is used, but this does not take into account that the warming rate increases with elevation (see some suggested references below). It is worth to briefly discuss it in this section.

Pepin N, Bradley RS, Diaz HF et al (2015) Elevation-dependent warming in mountain regions of the world. Nat Clim Chang 5:424–430. doi:10.1038/nclimate2563

Nigrelli, G., Fratianni, S., Zampollo, A., Turconi, L., & Chiarle, M. (2018). The altitudinal temperature lapse rates applied to high elevation rockfalls studies in the Western European Alps. Theoretical and Applied Climatology, 131, 1479-1491.

L 212 The Delaunay method assumes a smooth transition among points, but temperature gradients can be non-linear, especially in regions characterized by microclimates or highly varying topography (as raised in the previous point). The authors should be cautious of how elevation is included in the model, especially in mountain regions with high complex topography and when weather stations with high elevation difference are used. Moreover, this triangulation assumes uniformity in space, this means that stations should be distributed in a reasonably uniform manner. In the case of sparse station networks or if the stations are unevenly spaced, as often occur in complex terrain like mountain regions, the method could not accurately represent the spatial variation of the variables, leading to skewed results. Also, the sensitivity to outliers should be considered (that is, the fact that interpolation process could amplify these errors).

Figure 2 Define A and B

Results

L 239 Did the authors set a minimum record length in the period 1970-2019?

Figure 4-5 It is not clear to what trend the arrows refer to, please clarify.

L 315-316 “which delay summer and advances winter” sounds misleading, please rephrase.

L 356 Not sure the aim here is to assess a correlation in statistical sense.

L 364 Not clear if the authors are referring to total precipitation or precipitation intensity (also in Figure 11).

Discussion

Some key findings are presented in the Discussion rather than the Results section, which diminishes the clarity of both parts. The Results section should be dedicated solely to presenting the outcomes of the analysis. Outputs should be moved in the dedicated section, leaving comments and comparison to other works here. As an example, Section 5.1 Climate which addresses changing climate patterns and long-term trends over the last decades in the area is not suited to the Discussion section in its current form (additionally, a climatological introduction of the study area should be included in the Study area section). Similarly, some parts of the methodology are presented for the first time within this section, which contradicts its intended purpose.

L 506-519 The RAPS method is mentioned in the discussion for the first time, but it should be included in the methodology, since it supports results and conclusions of the work.

L 537 How did the authors measure the correlation?

Figure 17-20: these and related description are more suited for the results section (e.g., L 452-457, L 485-489).

Figure 21 As in the previous comment, results of this analysis (L 560 on) this should be anticipated in the Results section and briefly contextualized in the Discussion.

Technical corrections

L 1 Alpine areas are undergoing “a high change” in…

L 12 An anticipation of “both the onset of summer and the end of winter”…

L 15 over “the” last…

Figure 11a) Correct “conditional”

Figure 10 Please indicate the elevation ranges in d)

Review of the manuscript "Rockfall triggering and meteorological variables in the Dolomites (Italian Eastern Alps)" by Bonometti et al.

General comments

As stated in the introduction, the aim of the work is to calculate the frequency variation in time and space of different climate variables in the Italian Eastern Alps, to understand the climate evolution in the area and its influence on rockfall frequency distribution. The subject is of major interest for the understanding of rockfall failure mechanism and hazard assessment, and the manuscript represent a substantial contribution. The results are discussed in an appropriate way and compared to previous works on the same topic. I think the paper is well written.

However, the methods used should be better explained. That is why I believe that some major revisions are needed to enhance the overall quality and clarity of the paper before acceptance for publication. Also, some parts of the methodology are discussed in the Results and should be moved in the Methods section.

Minor corrections are suggested in the pdf and the more important points are developed hereafter. Some references of the text are not in the reference list. Please check that all the references are in the list.

Specific comments

Conditional probability (refers to the abstract and the Methods section)

Line 16. To be clearer the expression "to study conditional probability of meteorological variables on rockfall events" should be replaced by "to study the conditional probability of meteorological variables knowing that a rockfall event occurs" (use the academic formulation). But is this true?

I have a doubt because: (a) The equation (10) gives the conditional probability of a rockfall event knowing a meteorological variable is in a given range; (b) This sentence (line 235) "This study focuses on the effects of meteorological variables in triggering rockfall events" suggests that the results rather present the probability that rockfall events occur given that a meteorological variable is within a given range. This seems more useful.

In section 4.3 (Results), the sentences like " Figure 11 shows the conditional probabilities of cumulative rainfall obtained from weather stations below 1000masl" suggest the opposite. Please precise explicitly in the introduction of the section 4.3 what probabilities are presented in this section!

Terminology

The term "rainfall intensity" refers to the rate at which rain falls over a specific period and is expressed in mm/hour. In this paper, the variable considered is the "rainfall height" or "cumulated rainfall" or simply "rainfall". The term should be corrected unless you really want to speak about the rainfall intensity (in mm/hour).

Methods - Bayesian method

The authors should explain why they have to use the Bayesian method (equation 10). Why don't they directly calculate P(R/Mi) by dividing the number of days with rockfalls when the meteorological variable is within the range 𝑖, by the number of days when the meteorological variable is within the range 𝑖? The way of calculating the different probabilities of equation 10 should be explained.

Results

Rockfalls and climate variables (from line 355)

Line 357. Please explain (discuss) why a weather variable has a different effect according to the season or elevation. For example, has freezing different effects if it occurs in autumn or in winter? Has a high temperature different effects if it occurs in autumn or in spring?

Variations of rockfall probability according to the decade are pointed out, but the variations according to the weather factor range should also be commented too.

It would be interesting to compare the conditional probability P(R/Mi) with the rockfall probability P(R).

Line 410-413: Not clear. Equation 10 gives the expression of P(R/Mi) and not P(Mi/R). Please name explicitly the variables considered.

RAPS method

I suggest to complete the presentation of the RAPS method with this sentence: "A trend of the rainfall is suggested by a parabolic trend of the RAPS (downward parabola for an increase)". See the figure 3 in Garbrecht and Fernandez, and the comment in the same page. A trend on the RAPS plot must not be confused with a trend of the rainfall.

Lines 517-519. The trends mentioned by Garbrecht and Fernandez are trends on the RAPS plot, but not trends for the annual rainfall. They highlight a shift that was the result of the relocation of the station. After correction no trend for the rainfall is mentioned. So, I suggest to suppress these lines.

Also, to avoid any confusion, I suggest to modify the lines 520-526 as follows: " In this work, RAPS analysis by altitude, was performed considering the 12 meteorological stations (Figure 20). Below 1000m, from 1974 to 2001, an upward parabola on the RAPS plot shows a downward trend of the rainfall and a from 2005 to 2019, a down ward parabola on the RAPS plot shows an upward trend of the rainfall. In 2002, a sharp increase is noted, likely corresponding to high rainfall events in May and November (Bollettino meteorologico e valanghe, Ufficio idrografico di Bolzano; Protezione Civile Provincia Autonoma di Trento). Between 1000m and 2000m, the downward parabola shows an increase of rainfall for the whole period, which accelerates from 2005. Above 2000 m, the downward parabola from 2003 to 2019 shows an increase of rainfall for this period."

Figure 20. Equation 14 implies that RAPSn = 0, as can be seen in the figures of Garbrecht and Fernandez. Yet, it is not the case in Fig. 20 of the manuscript. Is there a error in the calculation?

Discussion (Rockfall)

Line 564-570. This paragraph is not clear. To be significant, the number of anomalies should be reported to the number of rockfalls in each season, range of elevation or range of volume.

I suggest to highlight this point: For rainfall, it appears that positive anomalies are much more frequently positive than negative (Fig. 21a-b), showing that large cumulated rainfall favours rockfall occurrence.

Figure 21. Do ST, LT, WT in the figure correspond to daily, weekly, monthly and quarterly in the text? This is not clear. Please use the same terms and give explicitly the aggregation scale in the legend. What is the difference between points and bars in Fig. 21 c and d?