thank you for considering my comments and revising the manuscript. While parts of the paper have improved, I think there are still some major issues (mainly on the wind-noise model and the presentation quality), which need be addressed. Please find my comments below.
1. Model for wind-induced seismic noise
In the revised version, you introduced a piecewise continuous function (eq. 1) to describe wind-induced noise. Is the function overall continuous, i.e. also at 6m/s wind speed? From the text, I conclude that the parameters for both function pieces are determined independently from each other via linear regression, in which case the two slopes will very likely intersect at a different wind speed than 6m/s (hence introducing an offset at 6m/s). However, from Fig. 3b, I conclude that the resulting function (red line) is continuous at 6m/s – is there some condition on the continuity of the function, i.e. y1(6m/s) = y2(6m/s)?
In addition, you state that the model is frequency dependent, yet, in the abstract you present values for the slopes in dB/m/s without any frequency context.
Also, I think it would be good to briefly state that the seismic power appears to scale to first order linearly with wind speed or something similar. In the current version, it sounds a bit that it would be a given that there is a linear relationship.
Finally, it is not clear to me, why the RMS amplitude of ground velocity calculated from the model increases exponentially with wind noise (Fig. 4b), whereas the acceleration PSD scales linearly with wind noise. Where does the exponential relationship come from?
Please clarify these issues.
2. Presentation quality
Most importantly, the spectrograms and fonts of Fig. 6 are way too small and make the figure unreadable. In addition, the quality of the manuscript‘s text would greatly benefit from a thorough proofread. Many parts are written imprecisely and suffer from grammatical errors. Some sentences are hardly understandable, e.g. „The wind speed used in this study is the average of the maximum wind speed recorded by the AWS which is measured at 2 m height every 10 minutes and averaged for each hour of 2017.“ Please revise the text.
3. I still think it might be worth to show the icequake detection rate as a function of the wind speed. In their response letter, the authors present a similar figure, but both quantities as a function of time. To stress the point that less icequakes are detected at high wind speeds, it would be better to plot the icequake rate (y-axis) directly as a function of the wind speed (x-axis), independent of time. This would stress the point that wind puts a bias on icequake detections and would strengthen the results of the paper.
- Introduction: give reference for the crevasse propagation velocity
- Sentences like „owned by ETH-Zurich“ and „map was drawn with QGIS“ should probably be moved to the acknowledgements section
- The formatting of equation 1 and its explanation need to be reworked e.g. say „… where x is the wind speed ...“ instead of „x = wind speed“
- In connection with Fig. 5, „cumulative“ number of icequakes is wrong, as it is actually an icequake rate
- „472 events were manually detected by Thierry Camelbeeck (Lombardi et al., 2019).“ It is not relevant here, who detected the events, delete!
- Fig3, caption: the last sentence does not make sense
- I think the term spectrogram is not correct. You are calculating noise levels as a function of frequency and wind speed, but that‘s not really a spectrogram (nor a seismogram, which is also mentioned once in the text, I believe).
- Table 1: the elevation of the automatic weather station is missing