Seismic evidences of the COVID-19 lockdown measures: Eastern Sicily case of study

During the COVID-19 pandemic, most countries worldwide put in place social interventions, consisting of restricting the mobility of citizens, aimed at slowing and mitigating the spread of the epidemic. In particular, Italy, as the first European country violently struck by the COVID-19 outbreak, applied a sequence of progressive restrictions to reduce both human mobility and human-to-human contacts from the end of February to mid-March 2020. Here, we analysed the seismic signatures of these lockdown measures in the densely populated Eastern Sicily, characterised by the presence of a permanent 15 seismic network used for both seismic and volcanic monitoring. We specifically emphasize how the amount of the amplitude reduction of anthropogenic seismic noise (reaching ~50-60%), its temporal pattern and spectral content are strongly stationdependent. As for the latter, we exhibited that on average the frequencies above 10 Hz are the most influenced by the anthropogenic seismic noise. Finally, we found an impressive similarity between the temporal patterns of anthropogenic seismic noise and human mobility, as quantified by the mobile phone-derived data shared by Google, Facebook and Apple. 20 These results further confirm how seismic data, routinely acquired worldwide for seismic and volcanic surveillance, can be used to monitor human mobility too.

This contribution describes the effect of quieting following the COVID19 lockdown measures on the noise level in a regional seismic network located around the Etna, Sicily, Italy. The subject is of interest, in particular in the framework of the "Social seismology" SE Special issue.
The paper is well-written, the structure is in general well shaped and the figures are of good quality (although many labels should be enlarged). Therefore, I think that the manuscript deserves to be published in SE after minor to moderate revision.
However, there are a number of point that, in my opinion, should be reworked in the final version of the manuscript.
The most valuable contribution of the manuscript is documenting that a seismic noise C1 reduction can be observed in areas far from large cities, where human activity still affect the background seismic noise via ship transit, touristic excursions etc.. Even some stations which seems to be installed at remote places reflect the decrease of activity following lockdown. I think that this point has to be highlighted through the manuscript and in particular in the abstract and conclusions.
In general, the manuscript makes a good job in presenting the data, but tends to be too concise in the interpretation part. Sections 2.2 and 2.3 are merely descriptive of the results presented in the corresponding figures. The reader has to wait till section 3 to learn something on the information included in the figures. I propose to include here the discussion on the differences observed between sites, the tentative origin of noise at each site etc. included now at Section 3. Figure 6 and 7 provide essentially the same information that Figs 2 and 3, presented in a different way. I will appreciate a comment on which are the advantages of each kind of representation. Are there features only observed these representations and not in the RMS or spectra?? If yes, it will be interesting to comment. Otherwise, the figures can be seen as redundant Regarding the comparison with mobility data, I think that the message that seismic data is consistent with other data is best passed using a graphic as that presented at Suppl. In this way the reader will first see an example of correlation between RMS and mobility data for specific station, then see the overall correlation and finally see the differences between stations and mobility data.
The discussion on Spearman correlation and t-test and p-values is unclear. I think that the original Figure 9 has to be used shown the station with good or poor correlation with mobility data, in some graphic, easy to interpretate way.
In the discussion (line 220) it is stated that only the ESAL/Facebook correlation does C2 not match the criteria, but, in my opinion, some of the stations (EFIU, HSRS, ESML, HPAC) clearly show a good correlation between seismic RMS and quietening, while for the rest, the correlation is less clear. This point should be clearly stated, noting that the relationship seismic noise/mobility is not always clear. In the Conclusions section it is correctly stated that the effect is strongly station-dependent. I think that this dependency should be better described here. In any case I would enhance the fact that, even for stations with poor correlation there are evidences of changes in the seismic noise values. Fig A3; P-values are in the order of 0.005 for Google, 0.0002 for Apple and 0.02 for Facebook. Are those order of magnitude differences realistic?? The authors state the p-value of 0.05 is considered sufficient to reject the null hypothesis; while this particular number is chosen?

I don't understand Suppl
Noise level variations related to ship activity or touristic excursions is interesting and not often described. I suggest to give more weight to this funny observation.
The section on the improvement on detection capability has a large potential interest, but it is not really developed here. In the main text, the authors just describe Figure  11 and the final discussion includes just a sentence on this subject. If the authors decide to keep the section, a significant improvement will be needed. Figure 11 shows that the number of pickings increase during lockdown, but the relevant information will be if more small magnitude events are detected or if the hypocentral determination is improved during lockdown. This analysis should be taken carefully, taking into account the epicentral distance of the events detected in each period, the occurrence of swarms/aftershocks that could perturb the comparison etc. Otherwise, a better option will be to keep the detection improvement discussion for a next paper focused on this subject.
Other points.
L. 31: Not sure that references to papers dealing with pharmacology are needed here C3 L. 65: The actual location setting of each location is hard to see in the small size screenshots in Supp Fig A1. I propose to summarize in this paragraph the different setting of the stations; how many are in towns, near roads, in small islands, in open nature etc Also a comment on the kind of installation used in each case will be useful; different installation types (vault, buried, building basement, insulation system etc) could affect the sensibility to human activity noise.
Line 85: The authors should explain why they decided to use the 10-30 Hz band. I suggest to use the submitted Figure 8 to justify this choice.