The manuscript submitted by Tunini et al. describes GNSS processing for station networks in Italy. While the topic is of general interest and fits the journal's scope, I see serious issues with the current draft. From a geodetic point of view, I like to ask the authors to revise the paper carefully. There are several issues with using geodetic terminology and common geodetic knowledge. Moreover, I found several inconsistencies and wording issues that have to be clarified. Finally, the authors report lengthy technical details clearly specified in different guidelines, these paragraphs should be rewritten in a much shorter way. Overall, I'm in favor of recommending the paper for publication after some general editing.

Detailed comments:

l10 please re-phrase "take the most from ..."

l25 please re-phrase "GNSS provides a", strictly spoken GNSS provides some navigatio signals to the users ....

l27 what is "GNSS technology"?

l30 what is a GNSS sensor, I guess you are referring to the antenna?

l31 you can determine sub-mm positions just from one day, I guess you are referring to time series? You might add clock corrections to the precise orbits

l40ff There is no need to explain why you processed the data individually. However, please cite NGL in a proper way and re-phrase "tune" by "customize"

l44 "beginning of the century" means "early 2000s"?

l57 XX century = 20 century?

l61 15-20km is the spatial resolution (or the station distance) but not the scale. "A more" sounds wrong?

l66 what do you consider as positions? Do you distinguish between coordinates and positions?

Fig 2 check the wording

l105 references for ITRF2014 and IGS are missing

l111 "tens of data" -> "tens of stations"?

Fig3 could you explain the drop in 2013?

l130 please shorten this part, these are common geodetic standards

l143 "log sheets" -> "sitelogs"

l149 Please rephrase "piece of information ..." + shorten this part, you don't need to report the IGS site guidelines (just cite them)

l202 language check for "that implements"

l212 you are using a Helmert blocking approach (clustering) which is very useful

l215 Honestly, I don't understand the issue with these old receivers. Are these cross-correlation receivers?

l221 in terms of orbits you should not write "fixed to ... values", they are introduced apriori (the fixing is obvious if you don't estimate orbits). However, a reference is missing here.

Tab1 2nd order ionosphere is corrected using the IONEX, please re-phrase; what's about other non-tidal loadings; I miss the orbit & clock products here; please add the estimated parameters and their characteristic here

l235 I guess this sentence is about the datum definition. Please explain clearly what was done and why.

l244 what's the meaning of "combining"? Are you stack normal equations?

l251 usually we call them "datum sites" or "datum stations", the term "stabilization sites" might be a bit outdated

Fig6 there are well-known assessments of computation time regarding number of stations, you might reference related publications

Fig7 what is "stabilization", if these are the "datum stations" you are going to constrain most of your stations. Is there any reason for doing this? Usually, you estimate stations in the area of investigation as free as possible ...

l297 what are stabilization errors?

l312 I don't understand the consideration of velocity biases. I don't think this is used in Blewitt and Lavallee, 2002. Why do you cite Masson et al 2019 in this regard?

l320ff We might have new stations but they are not young. Language check for "between others, we cite"

Fig 10 remove topography from these plots, are you showing vz (velocity in z direction)? Overall, you are assuming uncorrelated coordinates in the velocity determination?

l370 the difference between VMF1 and GMF was done by many colleagues, refer to their studies. It's kind of useless to skip ocean tide loading, I recommend to remove this part.

l391 What are "time changes in the environment"

l412 What is "variable atmopsheric noise"?

The paper illustrates the analysis strategy of GNSS data in the eastern portion of the Alpine arc, including a broad network of permanent stations. The procedures are well reported and the results are significant and valuable for tectonic studies and slow deforming processes investigations. All the outcomes are publicly available and well documented. I would strongly recommend the publication with only a few minor revisions and a general english editing to enhance readability.

Minor revisions:

Row 79-81: This sentence is rather generic and does not give details on your approach. I would suggest to either delete it or to better explain some terms, i.e. what kind of "slight bias" or "parameter" may compromise the quality (define which one). Or focus on your particular choices that should minimize  the error sources.

Row 198-199: It is not clear if the data processing considers only GPS observations or if it includes other constellations. Please detail this here.

Row 215: What does “old receiver” mean? Are they single frequency receivers? Please add more details.

Row 233 and 378: The Helmert transformation as proposed by the authors is rather peculiar, not alligned to EUREF standards in which only no-net-translations are imposed, please justify your choice.

Row 244-250: This paragraph should be revised, try to better clarify what steps you did to get velocities. e.g.

1) it seems that you combine (in time domain?) daily solutions to get velocities: this is not clear enough.

2) you state the need to estimate both rotations and rotation-rates independently of EOP. It is not clear what data do you use as input in the estimation process.

3) Is the velocity/position estimated using least squares or kalman filtering?

Figure 10: A few velocity vectors don't show the arrowhead, why? Please explain in the caption, at least.

I also uploaded the manuscript pdf file with a number of text editing corrections.