The article under review cover an interesting topic and present some unprecendently published results, for these reasons I would be in favour forpublication. Unfortunately, the manuscript does not fit the standard for publication and It requires, in my opinion: a deep review for different reasons:

• the text is in some part confuse, with a lot of repetitions and for the reader (at least for myself) it is difficult to distinguish between original results, hypothesis from the authors and previously published results. It emerges clearly that this is the manuscript from a newbie researcher and I would encourage him to rephrase many sentences and drain the text as much as possible, to ease the readability and comprehension.
• a clear description of the methodology is missing. I understand that PSD is a standard but how PSDs were computed should be described, otherwise the results would be difficult to be reproducible. The authors do not mention is disturbances to the noise (e.g. earthquakes) are removed to the dataset.
• Section results does not analyse in depth the results, e.g. figure 3 that contains the substance of the paper (PSD for single station, difference between the different months and years but it is not discussed at all.
• In Section Discussion I would suggest that the authors at first present their results and then they discuss them in the context of previous study. figures are difficult to be read. Italy is long and narrow and the authors are evaluating +500 stations that means +500 colored symbols  placed in the map.

We are at the end of 2022 and the authors during the review phase will have the full 2022 year available. I encourage them to use that dataset to provide a much comprehensive analyse for a complete year for which a lot of stations should be available. and eventually to consider the option of dropping data from 2019 that could become less relevant.

Moreover:

L.20 To complete the though I would suggest to add that in this case earthquakes are considered as disturbances in the signal.

L.21 Since the authors made a distinction, we now need a definition of what, for the case of this paper, is noise.

L.25 I would suggest to add also the scattering at shallow layers that e.g. generates the so-called Newtonian Noise (e.g. Harms et al. 2009.)

L42 “away from anthropogenic noises”, I would say “far from any source of noise”, usually seismometers are buried to prevent thermal fluctuation, and so on.

L.44 Since the argument is faced in a general perspective, I would say that seismic stations are placed where it is appropriate for the purpose of the project itself. VBB stations are in remote places far from anything, accelerometer for site effects and strong motion are placed at the study site and so on.

L.50 I would suggest to extend this sentence. It would be difficult to understand why pandemic reduced the noise. Eventually including the citation of some of the paper on this topic as Lecoqe et al, Piccinini et al, Poli et al.

L.54 “section 3 section” is a mistake

L.57 Please note that covid and COVID-19 are the same thing, same for “covid lockdown” and “COVID-19 lockdown”. please fix it using one name over the whole manuscript

L.62 at line 46 RAN was called in a different manner “Integrated italian Accelerometric Network” In my view things should be called consistently along the manuscript

L.63 At line 47 the contributors are differently described. If there is the need to repeat it, please be consistent.

L.65 “in the South” and “in the North East” please specify of what,  South of Italy I suppose.

L.66 I am getting confused, The authors use RAN as the acronym for the Integrated …., then they write that the RAN is made by three networks. And one of the three is the RAN.

L.68 Again there is some redundancy in the description, the fact that some of them have been converted to continuous was already mentioned about.

L.71 Third time the migration to continuos was mentioned.

L.73 at line 59, it is written that, for simplicity The authors will call it just lockdown.

L.75 Piccinini et al, proved that this was not true at national scale.

L.79 Question: data from 2021 would not be useful to integrate the dataset?

L.87 I would suggest to add a sentence describing the workflow to go from data (continuous time series) to PSD. e.g. data have been corrected for the response? How the spectrum was computes is not mentioned.

L90 I am not english mother tongue, but I feel that it is more appropriate to write “data” in place of “the data”. Please check.

L.104 better to say "described" if the author extend it, as suggested above, to a full description.

L.104-105 figure 3 is not discussed although it contains THE RESULTS of the analysis. The reader cannot understand where the considered few stations are located and why they differ in noise level.

L.105 This sentence is not clear, results are shown in fig.3, what is then in fig.4?

L.107 I am feeling pedant but is RAN stands for Rete Acceleromentrica Nazionale, then it is not necessary to follow it by network.

L.107-116 t the authors move from periods to frequency and backward. I understand that this is a common practice but, in a paper it is more appropriate to stick on one choice, otherwise the reader gets confused.

L.120 RAN stations at touristic sites can experience the opposite, quiet in the weekdays and noise in the weekend. Anthropic noise is very local. Did the authors consider it?

L.122 this is a repetition of line 119

L.123 english unclear to me

L.125 “very long period” please give the period band since for some seismologists this would be tens and even hundreds of seconds

L.133 Since the author proved that a seasonal variation and a weeday/weekend variation exist, I wonder if they considered it when comparing lockdown and no-lockdown. I mean that, to be consistent and to catch only the lockdown effect, the comparison should be done only with the same time span of 2019 and 2022.

L.140-143 Sentence is too vague

L.144 and following, Since the effect is local, did the authors consider the eventual presence of Wind Farms, or other facility that could produce anthropogenic noise at longer periods?

L.149 “is assumed” It is not an assumption, it is an observation from data and from road traffic data, mobility from mobile phone records and so on. There is plenty of data showing that human activity is reduced.

L.153 “trend” I would say pattern.

L.153 “of” typo?

L.155 “if a station is located in a settlement” I would expect that this is one of the result of this study, not and hypothesis within the discussion section. is this observed in data or not?

L.158-160 Again, do the authors observe what described by other authors in their analysis? This is not a review paper but a scientific one.

L.169 “stations start”. Start means that there is a variation to me, when do they start? No clear.

L.175 If the last sentence applies, that implies that stations are blind in this range of period. I do not understand why discussing the source of noise when in this frequency band accelerometers just measure the self noise of the instrument. Moreover the self noise can be computed and measured. It is not a matter of believing. Am I wrong?

L.179 Again the authors discuss something that they cannot observe. I suspect this depends on the fact that they are using accelerometers and D’alessandro et al. (2021) used velocimeters.

L.185 “period periods” repetition

LInes 185-189, in summary:

1. human activity dominate noise in this freq band.
2. high noise can be linked to activities
3. less human activities less noise.

Do we need a scientific study  and a paper to state this? Different is when this is a direct observation from data. but this is not what the authors write in these 2 sentences.

L.190-194:  Covid reduced human activity, ok.  Human activity influence seismic noise. Noise is higher in populated areas and near buildings.  A dozen of paper noted such a reduction. The authors too. OK, what is the added value of this study for the Covid-19 lockdown ? It is not clear to me.

L.215 “are” should be “is”

L.216 “at” should be “in”

L.222 “dates” ???? I presume median of the PSD noise.

L.223 “are more dominant” could be “prevail”?

L.233 I miss to understand how this paragraph, at the end of section discussion is linked to the rest of the study. It would make sense at the beginning of the analysis when authors tackle the problem of distinguising between different source of noise and of characterize their frequency content.

L.235-237 I suggest to rephrase the sentence.

L.240 “manually” “by hand”, As I mentioned, I am not english mother tongue. But to me, this sentence means that somebody was checking the passage of cars using his own hands. Not that, as I suppose, somebody visually inspected seismic data and searched for the effect of the  passage of the cars and manually marked it on the seismic trace.

L.247 “is” should be “are”

L.254 the assertion “capable of providing ….” Is a qualitative speculation not based on true values.  can the authors give some estimate of the miminum magnitude that can be detected at local distance by high noise and normal noise accelerometric stations?

L.256 “…but also the small ones” This sounds a bit obvious and not very useful without, as above, an estimate of the detection capability. Big and small are always relative to something.

L.258 “they” I cannot understand who is the subject: Selection criterion for what?

L.259 “Some of the stations” How many? again description of data and of result is too vague for a scientific paper

L.260 “(528…. ) whereas some of them” In the data description it is written that the study is based on 528 stations. If 528 are in settlement, how can be that some of them are away from settlement?

L.263 “in the short period” could be “in the short period band”?

L.273 and following. This is a repetiition of line 172 and following

L.281 “.. is applied” not clear

L.296 Anthony et al 2021 was published in 2022.

Figure 1, I was surprised to see that Anthony et al. (2021, actually 2022) report info for only such narrow band and I checked the paper where for example I read (last paragraph, second column, pag 648) that noise in the band 0.0625-1 second contains cultural noise. So the narrow band should be as large as covering the entire band. Please check also the other.

Figure 2a, by placing the closeup box over Sardinia, the reader misses to appreciate the network coverage in that portion of the study area that is the whole Italian country.

Figure 2, I wonder if there is a reason to plot stations with reverse triangles while in Figure 4 are not.

Figure 2 caption, the color coding of the figures is not descriptor. Moreover I do not understand what (RF) stands for.

Figure 2 palette:I I read PSD database ompleteness  does this means that the authors counted the expected number of PSD for e complete time-series and then computed the ration of available ones?

Figure 3, caption says “several stations” while in the manuscript I read few stations and actually there are a very small fraction of 528.

Figure 4, in the caption the authors use “Power Change” while in the caption and at line 113 I read “noise”. Since “power change” is introduced in the discussion and not in the caption I do not understand what figure 4 display.

Figure 8, seasonal variability. In the manuscript it is mentioned that: 1) seasonal variability is studied only for year 2019 and it makes sense since data coverage for 2022 is limited to January-April. It is also mentioned that data analysis is limited to stations with completeness above 90% and it also make sense. In figure 8 I see only two triangle in Pianura Padana and one of them in figure 2a is colored in green that means ~40% to my understanding. Apparently something is not correct. I wonder is this applies also to other stations.

Figure 12, the authors did not provide indication of where Trieste is.

Figure 13, If I interpet correctly this figure line for 00:45 ± 45 (purple with dots) has high noise at .1 seconds, while line for 23:15 ± 45 (red without dots ) has low noise. How can be midnight much noiser than 11pm? This contractics expectation described in the manuscript. Am I wrong?

Figure 14, “demonstrated” better to say locatized?

Table 1, It confuses me. Since the total of stations gives 715 but the authors used only 528 of them. Is it revenant this table ?

The manuscript by Fornasari et al. is original and intriguing because it faces different issues, among others the quality of the seismic networks and the origin of the seismic noise. Moreover, it shows the effect of the COVID-19 lockdown in Italy on noise, which was expected but never seen in consideration of the unicity of the lockdown period. 

Nevertheless, the writing and the general organization of the work are critical. In my opinion this manuscript needs major revisions to be published. What I found to be critical in general is that the writing is really confusing, there are lots of repetitions, refuses, some useless or appended sentences and different jumps in the order of the communication. The result is that the reader cannot understand what the authors want to say. I would encourage the authors in publishing (especially if newbie researchers as at least the main author is) with a point-to-point revision of the text, but sorry I have no time to do that. However, I try to point out and suggest something as follows:

• the stations used for the analysis are not very well presented. The authors use RAN for the real RAN but also for RAF and ISNet. I suggest to be very clear in the paper or to figure out something that does not actually exist, a sort of Integrated National Accelerometric Network (INAN). I don’t know if this latter it’s a good strategy: my concern is that we are not considering the INGV accelerometric stations, so we cannot define “National” the integrated network. By the way, why didn't the authors consider the INGV stations?
• Also the number of used stations is not very clear. RAN consists of 647 digital stations, RAF of 14 stations and ISNet of 31 stations. Overall, there are 692 accelerometers, no “more of 700” as reported in line #63. The authors say that they used 528 stations because many of them still run in trigger acquisition mode and this number changed over time. So, what should be clear is: how many stations did the authors use in 2019, 2020 and 2022 for the analysis? Figure 2 is difficult to read (by the way, what do the different colors mean in the Figure?) and it is poorly commented.
• Table 1 is also confusing about the stations actually used (715 stations are reported) and in my opinion is useless. A sentence in the text is enough to describe the deployments.
• the paper focuses on the accelerometric stations and the authors show several results for different frequencies (periods). The authors should mention that the accelerometers are not very sensible to low motion, such as noise, especially at low frequencies. Then it’s normal that the low-frequency noise recorded by the stations could be out of meaning.
• the method of computation is merely cited and the differences between the real used method and the standard ones are only reported in Table 2, but not too much discussed. In my opinion the authors should at least present the formula of PSD, the data preprocessing (e.g. instrumental correction, the kind of spectra) but also clearly explain their choices. In other words, the authors should answer these questions: what are the improvements using longer windows and the linear interpolation for gaps? Why don't they use the standard computation for PSD?
  
  Another point is that there is no word on how the authors took into account the earthquakes or other strong transients that occurred in the time series. Have they been cut off or maintained? 
• the Results Section is not very well explained and organized. Figure 3 and then Figure 4 are presented as “the representative noise level” of some (or all) stations. Then there is a (very short) discussion about the day-night, weekend and seasonal variations of noise, variations that can be important. So, I am a little bit confused about what the authors consider “representative”: is it the night level, is the weekend level, or what?
  
  Figure 4 is the most important one, and in my opinion deserves more discussion. Unfortunately the figure (and this is the same for all the figures like this) is very hard to be read. In this case, I suggest extracting from Table S1 the first and last 10 stations ordered by the noise level, it could help in the interpretation. 
  
  Another point is Figure 5. What is the meaning of it? In my opinion it’s a general overview of the entire “integrated” network but it’s not very useful to know that the network is “good” in average.
  
  Figure 6 and 7: it’s not very clear if the colour scale represents (as for Figure 8) the difference in noise levels between supposed calm (night, weekend) and disturbed (day, weekday, winter) periods.
  
  Moreover, the authors present the PSD analysis for the COVID-19 lockdown period, but it’s merely a list of Figures, without comments on them.
  
  In general I think that in this section the authors must clearly illustrate and comment on the results.
• The Discussion Section (also the others but this in particular) requires a deep English revision.
  
  The number of analysed “noise levels” is reported as 525(!).
  
  In the discussion about the sea, swells and/or wind effects on noise, the authors should take into account that the accelerometers are not the best kind of sensor to record these low frequencies (high periods).
  
  Lines #173-175 are a good example of what I mean with “jumps in the order of communication” and/or appended sentences. In the long period the accelerometers are completely deaf.
  
  Table 4 refers to a particular station? Or is an average?
  
  “Changes in the daytime are more significant than the changes in the nighttime between the lockdown and no-lockdown time span.” How do you explain that if the working activity was almost zero during the lockdown?
  
  I think that many comments should be moved to the Results Section.
  
  I don’t really understand the utility of introducing here sub-section 5.2. Maybe, but I am not sure, it could be moved in the Result Section after or together with sub-section 4.1.
  
  Also sub-section 5.3 has nothing to do with the Discussion Section, maybe it can be inserted in the Result Section but the authors should introduce the problem.
• What is missing is some consideration about the general quality of the sites of deployments of accelerometers. On the basis of the results of this paper, what are the effects on the strong-motion monitoring in Italy? I know that it’s a very wide answer, but an effort to answer should be done.

I found the “Scientific Significance”, “Scientific Quality” and “Presentation Quality”; Good.

The paper addresses relevant scientific and/or technical questions within the scope of NHESS, including new data and results which are presented in international standards.

The scientific methods and assumptions are valid and outlined clearly. The results are sufficient to support the interpretations and the conclusions. The title of the manuscript clearly and unambiguously reflect the paper's contents. The abstract provides a concise, complete and unambiguous summary of the work done and the results obtained. The title and the abstract are pertinent and easy to understand for a wide and diversified audience. The size, quality and readability of the figures are adequate for the type and quantity of data presented. Authors give proper credit to previous related work, and they indicate their contribution, clearly. The overall presentation is well structured, clear and easy to understand by a wide and general audience. The length of the paper is quite adequate; not too long, not too short.

The fluency of the paper is good, in general. I noticed some ambiguity in some of the statements which can be edited quickly and easily. You can kindly find my comments/corrections/suggestions which will improve and strengthen the submitted paper, I believe. I congratulate the writers and wish them success in their future work. Last but not least understanding the background noise levels and seismic network standards is very important for studies in earthquake locations and early warning systems. I also want to thank the journal editor for evaluating this submission which will be quite beneficial and enlightening for the data users of this seismic network.

This study is about the computation of the variations in ambient noise levels of the big Italian Strong Motion Network to evaluate the performance of the stations. The study is definitely important and useful for future studies. I appreciate the authors used a large dataset to establish the study by considering different periods and taking the advantage of COVID-19 lockdown period. The methods they used is a well-known and suited to the paper. The manuscript is written in good English. Although I like the idea of the paper which provides an excellent opportunity to exploit this large dataset for different time periods, I believe the paper still needs some significant revision. Please see my comments below.

General comments:

• The paper sometimes is lacking in quantification for the validation of the results appropriately, especially in the result section. For instance, I am a bit surprised there is no quantitative comparison with the results from the other networks around the world to evaluate the performance of the Italian Network in the Discussion part. I would definitely add one paragraph to the Introduction part, showing the previous studies and their ambient noise level with numbers, and compare&discuss them in the Discussion section.
• The paper also needs additions and extra explanations because some details are missing (e.g., in the Method section).
• The organization of the paper is not well structured. The aim of the study is not given clearly. The sections sometimes don’t show their actual points. While the Results section is very smooth, the Discussion part contains mostly the results of the study. Furthermore, the text sometimes contains repetitive sentences specifically in the Result, Discussion, and Conclusion parts. This does not make the article fully comprehensible.
• For the figures, the figure axis fonts are quite small and not readable. On the other hand, the authors can do zoom-in maps based on different coordinates instead of showing the whole land of Italy which makes the figures more catchy.

Minor comments:

Abstract

• Line 5: ...anthropogenic ...Please be consistent to write this term in the same way throughout the text also figure captions. For example, it is written as “anthropic” in Line 98.

Introduction

• Line 49: ...RAN accelerometric network… Only “RAN” is enough here. Please be consistent with the abbreviations throughout the paper.
• Line 56: ...covid lockdown… Please be consistent (e.g. line 57: COVID-19)
• Line 62: ...more than 700 stations… How many exactly?
• Can authors add also a paragraph from previous studies and their findings that use the same or different methods? Please also describe clearly the aim and purpose of your study which is missing in the text.

Data

• Please add a diagram to clearly show the sub-networks that are involved in National Accelerometric Network (A supplementary figure is fine).
• Could authors give more information about the stations (e.g., type of instruments, sensor, the cut-off frequency) used in their analysis?

Method

• This section lacks some technical points and needs some formulations. I was wondering if the authors could elaborate giving more information about the parameters and explaining more clearly why they were chosen and according to which criteria.

Results

• Line 105: ...0.1 s, 0.25 s, 0.5 s, 1 s, 2 s, and 5 s...Why did the authors choose these periods? Did they do any analysis to determine them? A spectrogram plot would be helpful for selected stations in order to understand dominant periods. Please explain it in more detail in the Data&Method section, not in the Result section.
• Line 125: ...The results show that winters are noisier than the summers...Can authors quantify how much (with numbers) ?
• Line 129 – 138. I don’t see any indication of results in these two paragraphs. Can you specify the ambient noise levels for the different periods that you selected for the lockdown period? Authors should mention the result in the text as they plotted in the figures in this section. If the authors don’t mention any results, the first paragraph rather seems related to the Introduction part while the second one is data and/or method related.

Discussion

• Please see my general comments above.
• Line 190-194. Did the authors compare their results with the studies mentioned in this paragraph during the lockdown period?

Conclusions

• Please avoid redundant sentences here and emphasize the main findings, contribution, and significance of your work.

Figures

• Figure 1: I am not sure if this figure should be included in the main text. It can go to the Supp. Material. Please see my general comments about the figures above.

Lastly, I believe the paper could be improved in many ways with some additions and restructuring as suggested above and published in the journal NHESS after applying the required revision.