The paper “Ground-based observations of periodic temperature fluctuations in the mesopause region with periods larger than 2 days” by C Kalicinsky et al., utilized more than 30 years of mesospheric temperature observations to study the oscillations and their long-term variations.  Lomb-Scargle periodogram (LSP) is used to identify the periods of these oscillations.  Fluctuations with periods of 5/6 days, 8-12 days, 15 days and 28 days were identified and related to Rossby waves.  Most of the activities occurred during the winter season for longer period waves, while the short period waves peaked during equinoxes.  The authors claim that long-term variations of the wave activity showed a quasi-bidecadal signature. This is a very valuable study and could provide crucial information about planetary waves and their long-term behavior.  However, I have some concerns that need to be addressed.

Major concerns:

1. I can’t agree with the way the authors counting wave events. By doing this, it makes the results in Figure 3 misleading and hard to understand.  I could not tell how many times the waves in each bin had happened during this >30 years period.  For example, the 28 day wave event claimed to be the most popular one, it has the same count as the 45-60 day waves, 350.  What does it mean? Which one occurred more often by how many times?  By looking at the duration of the 28 day wave in Figure 7 (~70 days), does it mean only ~5 times this wave occurred during the >30 years of observation?  From the spectrum results of Figure 2 and Figure 7, one can see that within one year, there were not a lot of wave events identified which is totally normal.

It is more reasonable to count each event as one which will clearly show how many times each wave happened throughout this very long data set.

1. The period range in Figure 4 is confusing with each subset inclusive of the others. In the top row, the difference between period >10 days and period >20 days are small, which indicates the waves with periods between 10 and 20 days, does not occur a lot. Does this mean the quasi-16 day waves only show up in your data very occasionally?  Again, the counting mechanism make it hard to understand the seasonal variations of the detected wave events.  Independent spectrum range would help with the results.
2. Section 3.2 focused more on the similarities and differences among the 3 quantities. It is not clear how introducing the 2 new proxies, especially the later one would help in drawing a clear conclusion.  In Figure 6b, only the ~20 year variation was mentioned, what about the shorter periods with even stronger power? How significant of this 20 year period oscillation?  What is the confidence level for this result? 
3. The LSP results for 28 day waves in Figure 7b does not make sense comparing with the data in Figure 7a. During the event of the 28 day wave, the spectrum results showed a very broad peak (period extended from 25 to nearly 50 days) while the data (Figure 7a) showed a highly defined wave period which would be a narrow horizontal maximum in Figure 7b. With such a broad peak, how the period of 28 days is concluded?  The other thing of this 28 day wave event is it happened in July and August, which is summer for the northern hemisphere which is contradict to what the authors conclusion (line 240).
4. In general, observations from one ground-based site are not enough for planetary wave mode identification. Section 4.1 tried to relate the observed periods of oscillations to known Rossby waves of certain mode. The Rossby mode has certain latitudinal and longitudinal structures. Relating waves observed from different latitudes can lead to wrong conclusions.  Also, for the waves of 4-6 days, 16 days, most of the studies cited were using observations of the mesospheric wind.  The normal modes of winds are quite different from the ones for temperatures.

Minor comments:

Line 100: please remove “by” at the beginning of the sentence.

Line 103: please remove "also”

Line 108: what is the meaning of the 01.07-30.06 mean? Dates? July 1^st to June 30^th (of the next year?)

Line 143: “their” mean?

Line 184 and later:  “shorter” period            

Line 357: “cannot be observed”, do the authors mean “was not observed “or the method is not capable of detecting it?

General Comments

This manuscript presents an analysis of seasonal and long-term periodicities in 30 years of mesopause region temperatures derived from hydroxyl (3-1) band spectrometer observations from Wuppertal, Germany.

They use a variable sized, sliding window to apply a Lomb Scargle periodogram analysis to deseasonalized time series for each year.

They analyse the main seasonal periodicities in the range of 2 to 60 days in terms of planetary Rossby wave activity. The majority of wave activity is observed in winter, with some indication of a maximum occurrence of shorter period waves at the equinoxes.  

Long-Term behaviour of wave activity was analysed using the standard deviation of temperature residuals and the mean amplitude of significant events as proxies. This revealed a quasi-bidecadal oscillation characterised by events with larger amplitude.

The work is a worthwhile study on planetary wave activity in one of the longest hydroxyl rotational temperature data sets and yields some sensible but not unexpected results in terms of predominant wave activity. Some further work and revisions are suggested below.

Specific comments

Line 14 Introduction section.

In the context of Lomb Scargle analyses of long term hydroxyl rotational data sets the work from the Davis, Antarctica observatory, seems particularly relevant (for example French and Klekociuk (2011) https://doi.org/10.1029/2011jd015731 , French et al (2020)  https://doi.org/10.5194/acp-20-8691-2020) but is not considered in the introduction or in the discussion on periodic fluctuations in 4.1. Including a comparison with these studies would be valuable, particularly since similar periodicities from other Antarctic sites are already discussed in the periodic‑fluctuations section (eg line 279).

Line 85 is there much variation in the amplitude and phase of the seasonal fit between each year? What are the ranges of the fit coefficients ?. What is the difference to a seasonal fit to all years ?

Line 100-105 why not do a sliding window LS over the entire deseasonalized time series instead of adding 2 months to either side of each year?. This paragraph also needs revising eg “all possible time windows have been of the complete time series were analysed” makes no sense.

The paragraph explaining the variable window length (lines 117-123) is incoherent and difficult to read and interpret. This should be revised for clarity.

The “factor” introduced in line 126 .. should be defined as cycles per window length of the period being analysed. How is this used in the analysis except to step up the window length for increasingly long periods analysed ?.   

Line 163 What is the criteria for significance here and in Fig 2 and how is this computation affected by the varying window lengths ?

Section 3.1 and Fig 3. Occurrence frequencies are counted by the centre days which meet the significance criteria. This will bias occurrence statistics to waves of longer duration. The occurrence of long period waves will necessarily occur on a larger number of days (and this is stated in Line 172). I don’t think this is a valid assessment of wave occurrence statistics and the analysis needs to be revised. Normalise by the period length or count waves of each period by event.

Figure 4. This is a confusing plot as the period ranges overlap. Why not separate these ranges and do periods <10, 10-20, 20-30 and >30 ?. 

Line 192 and Fig 5. In the discussion of observed amplitudes what is the effect of the integration of the temperature variation by a wave propagating through the ~8km layer profile of the hydroxyl emission ?. Does this limit the observable frequency and amplitude of waves.   

Section 3.2 and Figure 6.  In the discussion of long-term wave activity the clearly dominant periodicity in the weighted sum power spectrum at ~4 years is not discussed. This has higher power than any other measure including the bi-decadal oscillation. What is this attributed to and does it align with the quasi-quadrenial oscillation of French et al (2020) https://doi.org/10.5194/acp-20-8691-2020?.  

Corrections

Lines 11-13 are unclear -> revise 

Line 19 proofed -> proved

Line 24 an -> a

Line 31 large opportunity -> advantage 

Line 31 since -> for

Line 32 instrument can be maintained all the time -> instruments can be maintained continuously 

Line 35 exhibit -> provide 

Line 44 a much longer -> an extended 

Line 45 omit "which is"

Line 46 omit "the"

Line 47 omit "the"

Line 47 get rid of -> omit 

Line 59 intensively described in -> detailed in 

Line 63 since -> to

Line 64 now -> present 

Line 70 line -> lines 

Line 72 are given in -> occur on 

Line 75 vibrational -> rotational 

Line 110 has also -> also has a 

Line 115 during the -> in this 

Line 118 minimum length of the used window has -> minimum window length also has 

Line 126 relation -> relationship

Line 127 value can be seen as a factor that defines the -> is

Line 135 omit used 

Line 142 dd -> d

Line 150 less -> lower and better -> higher 

Line 152 omit large 

Line 153 (here in figure 2a) in brackets 

Line 163 minimal -> minimum 

Line 184 largest part of the events -> largest number of events 

Line 208 as last proxy -> a second proxy 

Line 209 the mean was subtracted before -> with mean subtracted 

Line 229 larger number -> the majority 

Line 238 which -> who 

Line 242 way -> considerably 

Line 244 omit different 

Line 248 the studies before -> previous studies 

Line 250 the results before -> the previous results 

Line 252 likeliest -> most likely 

Line 273 observe -> see

Line 282 activity is different for different type of waves -> activity varies for different types of waves 

Line 297 mechanism -> mechanisms (2 of) 

Line 300 omit "of the" 

Line 301 insert the difference 

Line 307 mechanism -> mechanisms 

Line 325 proceeds still -> continues 

Line 338 insert 'the' mesosphere 

Line 342 insert 'and' is beyond 

Line 354 remove apostrophe

Line 346 last decades -> this era