General Comments:

The paper compares several snow stability metrics derived from snowpack modeling against a rich dataset of field measurements collected at the Steintälli study site during the winter 2015-16. The stability metrics may be classed as process-based (skier stability index SK38, critical cut length r_c) or data-driven (machine learning snow instability metric P_unstable). The paper discusses the challenges of determining which weak layer in simulated stratigraphy output is the most critical and therefore the most likely to cause avalanche activity. The authors consider three approaches to selecting most critical weak layer: maximum P_unstable, relative threshold analysis (RTA), and persistent avalanche problem (pAp) approach. In total, six methods that perform the dual tasks of selecting the critical weak layer and quantifying its stability were evaluated, along with the non-modelled, three-day new snow (HN3d) benchmark.

The field measurement dataset is unusually detailed, comprising shear frame measurements, stability tests (PSTs, ECTs, CTs), detailed snow profiles, automated weather station data, and snow micro-penetrometer profiles. This, combined with how the season evolved to develop several persistent weak layers and several significant avalanche cycles, makes it an extremely valuable dataset, well suited to the task of comparing modeled stability metrics.

The manuscript is well written and the analysis is sound. The authors discuss limitations (training data/validation data overlap, and scale mismatch between point observations and regional danger ratings) and the results generally support the interpretations and conclusions. I recommend publication after the following comments are addressed.

Specific Comments:

The overall accuracies of the stability metrics reported in Table 5 suggest a range between 53% and 73% (PC of 0.53 to 0.73). Can you explain why the overall accuracy range suggested in the abstract is stated as 60-70%?

Several of the metrics have performance skill scores that are very similar, yet no formal confidence intervals are reported. Despite this, there are several places in the manuscript where the relative ranking of the metrics is discussed, admittedly with cautious language: e.g. L. 495 “P_unstable performed slightly better than the traditional skier stability index RTA-sk38”. How confident are you in their ranking? How confident would you be that the relative ranking would change if applied to a different area, or a different set of conditions in the same area?

Further to the similarity of some of the metrics, the authors explicitly mention (L. 418) that the strong negative correlation between P_unstable and pAp-r_c is expected as r_c is one of the 6 features considered to estimate P_unstable. Does this erode the statement in the conclusion that a multi-model approach may be advantageous for operational forecasting? Are there additional stability metrics available or being developed that offer a truly independent assessment from existing approaches?

The authors are honest about the overlap between the training dataset and the validation dataset (L. 453), where they state that the training dataset for P_unstable contained 11 profiles from Davos from 2015/16. Ideally the impact should be characterized by reporting whether excluding the 11 overlapping profiles from the training dataset changes the results. This feels especially pertinent in light of the discussion above, that relatively small differences in performance skill scores are being discussed and characterized. If such reanalysis is not possible, it would be better to highlight this limitation in a stronger way, perhaps by repeating it in the conclusion.

What is the rationale for correcting the avalanche danger level (L. 190)? If the uncorrected danger level was what was available to the public when the regional forecast was issued, isn’t that perhaps a more meaningful comparison, as it recognizes the limitations of human forecasting. As presented, the danger level has been altered to improve correlation with the AAI. Both are presented in Figure 4, ostensibly to offer comparisons to two different measures of avalanche danger. However, these are no longer independent, as one has been corrected against the other. I note that the caption describing the avalanche danger level (“Avalanche danger level for the region of Davos”) does not indicate that a corrected version of the danger level has been used. At a minimum, I would like to see the caption of Figure 4 updated to indicate that the corrected avalanche danger level is used.

PON (L. 334) is not defined (presumably the probability of non-events).

Figure 4: There is a small orange spot on Pane (c) in mid-March that does not have a corresponding orange shaded region on the lower half of Pane (a). I suggest adding a small shaded region on the P_unstable region of Pane (a).

Technical Corrections:

1. L. 146: “Fig.1” should read “Fig. 1” (with space)
2. L. 196: “Fig, 1” should read “Fig. 1” (period, not comma)
3. L. 225, Figure 2, x-axis label: “Winter 2025/16” should read “Winter 2015/16”
4. L. 252: “PWL’s” should read “PWLs”. Same comment for L. 400
5. L. 262: “10 March 2015” should read “10 March 2016”
6. L. 315: “Tab. 5” is referenced in two consecutive sentences. Delete the second instance.
7. L. 399: “4.2.1 Stability metrics” should read “4.2.2 Stability metrics” (incorrect section numbering)

The authors present a detailed study of snow stability metrics during a winter season in Davos, Switzerland that had multiple persistent weak layers and cycles of avalanche activity. The data was centered at the Steintälli study site above Davos, Switzerland which included stability tests, image analysis of propagation saw tests, and snow-micropenetrometer observations. Using data from the study site’s automated weather station, they ran the snow cover model SNOWPACK to generate simulated stratigraphy and modelled stability metrics. In their analysis they include information in the broader region via the avalanche activity index (AAI) and the daily public avalanche danger forecast.

They provide detailed comparisons of weak layer identification, weak layer properties (shear strength, density, fracture toughness), and stability metrics. Six stability metrics are included in the analysis.

They found general agreement between all six stability metrics which coincided with periods of higher AAI and avalanche danger. They provided some nuanced comparisons such as the ML approach, P_unstable, and combined skier stability index and critical cut length (pAp sk38 intersection with r_c) performing slightly better than the traditional skier stability index (RTA-sk38).

In the introduction, they provide a particularly well-written overview of the history, various attempts, and state of the science for different approaches to quantify snow stability. Their methods are generally clear, organized, well-referenced, and justified. Their conclusions follow well from the thorough results and discussion sections. The richness of the dataset is particularly valuable.

I recommend publication, after consideration of the following comments and technical notes:

• The authors state the dataset from Steintälli 2015/2016 provides a valuable benchmark for the evaluation of stability metrics and that the “wealth of data makes it an attractive case study” (lines 476-477). Thus, the title of the article may be too broad for the article's content, as it could, perhaps, be considered more of a case study. Consider adjusting the title accordingly.
• The very high AAI on 9 February 2016 that only overlapped with one or two (I can’t quite tell) stability metrics jumped out in Figure 4 but was only briefly mentioned in results and discussed some in lines 406-411. Consider drawing more attention to this result. I find it remarkable that the highest AAI day, by far, was missed by most stability metrics.
• Lines 107-108: Why is the propagation saw test not considered a stability test?
• Figure 1: Text in upper left and grain form symbols are not legible when printed.
• Lines 209-210: Are you basing this statement off higher number of taps for CT’s and ECT’s? If so, there is first a decrease, then, a subtle increase but the trend seems rather weak and based off N=1 for each test day.
• Figure 2: A bit confusing to see the same data plotted twice for shear strength.
• Figure 4: This figure synthesizes the major results from the analysis and could have improved clarity. For example, the stability metrics themselves could be labels for the vertical axis in (c). There is not a clear difference in labeling between the skier stability index in (a) and (b). There are borders in the shaded regions in (a) but not (b).
• Table 5: Consider restating the PC, POD, and PSS acronym definitions in the caption.
• Line 334: I can’t find a definition of the acronym PON.