General remarks:

This paper presents the development and validation of an automated pump efficiency measuring system for ozonesondes using the airbag method. An important topic in ozonesonde research to correct ozonesonde readings properly and currently also to get a better understanding of the recently observed total column ozone drop by ozonesondes (Stauffer et al., 2020). The paper describes the development and validation of the new system in all its technical and scientific aspects. After a short functional description of the ECC-sonde a technical overview of the setup of the measuring system is given, followed by a detailed description of the airbag methodology used to measure the pump efficiency at ambient air pressures between surface downto 3 hPa. The investigators also investigated and corrected for (i) back pressure effects; (ii) heat generation in the pump; (iii) differential pressure effects between airbag and ambient air in the vacuum desiccator; changes in the airbag capacity due to temperature changes. Since 2009 the automated system has been operationally used at three JMA sounding stations (Sapporo, Tateno, and Naha) for each flown ECC-ozonesonde. The results have been compared with pump efficiency measurements done by other investigators, e.g. Johnson et al. (2002; CMDL (bubble flow meter) and UWYO (bag method)) The authors have investigated the long term stability of the system to be better than a few percent per decade. In addition, a detailed statistical analysis of the time series of the pump efficiencies measured for the period 2009-2022 at the three JMA stations has been investigated and discussed in how far the fluctuations of the measured pump efficiencies observed between the different serial numbers of the sonde-pump can explain the total column ozone (TCO) drop of ECC sondes detected by Stauffer et al. (2020).

The paper is well structerized and written, whereby it is recommended that a native english speaking person should improve some of the english in the paper. However, the content of the paper in all its aspects is very good and tec hnically as well as scientifically. The presented techniques, methodologies, correction algorithms, measured time series of pump efficiencies, statistical results, and conclusions are new and are based on extensive and solid laboratory work and scientific analysis. All figures, tables and their layouts are appropriate to the results presented. The level of detail on the different corrections made and the observed variability among the time series of the different stations has been treated in an appropiate and balanced way. The paper is an important contribution to the ozonesonde community to fill in the existing gap of having only a few pump efficiency measurements between 2009 and 2022. The paper is certainly a milestone in ozonesonde research. Therefore, the paper fits very well in the scope of the Atmosphere Measurement Techniques, and I rate the paper as very good and recommend publication after only minor revisions as listed below.

Some comments

L10      Replace: emulating by simulating

L14      Replace: accumulated by collected

L24      Replace: the detailed by a detailed

L25      Replace: flew up by flown

L27      Replace: chemical by electrochemical

L31      Replace: 80% by more than 90%

L62      Replace: research themes in the past by other investigators

L63/64 Replace silicone membrane by bubble

L66      Replace:           the ECC- by ECC-

L124    Replace: to a small by of a small

Figure 8: Larger charactersize in legends

To be clear to the reader that after section 3-3 all further pump efficiencies reported in chapters 4 and 5 are always measured and determined with a 3 ml sensing solution in the cathode cell.

Figure 16: Larger charactersize of axis

This paper is an overview and summary of the long-term JMA ozonesonde pump flow efficiency measurements of individual ozonesondes launched at three JMA field sites. The standard ozonesonde procedure (SOP) includes several steps, including measuring flow rate of the ozonesonde at the surface during the day-of-flight preparation. The additional measurement of pump efficiency is not included in the SOP due to the complicated and very time-consuming nature of these flow rate measurements done at a range of low pressures from surface to 3hPa in a vacuum chamber. Thus, essentially all ozonesonde operations (except for JMA) around the globe use an average pump efficiency to calculate the ozone profile.  The average pump efficiencies are based on measurements and tests performed many years to decades ago.   The ozonesonde equation (converting current to ozone concentration) includes a term for the pressure-dependent pump efficiency or the drop off in the surface measured flow rate during ozonesonde preparation.  The authors also make an excellent point in the conclusions on the importance of long-term monitoring of pump efficiency of ozonesonde pumps from the two ECC ozonesonde manufacturers. 

The JMA designers of this ozonesonde flow rate calibration/efficiency method have done an excellent job in developing the pump flow measurement system and addressing and reducing  errors and offsets related to making low pressure flow measurements.  This includes automation of the key steps, measuring pump motor speed (tachometer), checking temperature effects and back pressure (cell solution) effects, and focusing on the plastic deflation/inflation bag to reduce restrictions by using a polyethylene bag with a thin fluoroplastic film placed inside the bag to prevent wrinkles during inflation and deflation. They also  simulated head pressure of the 3cc of sensor solution (the normal KI salt sensor water solution would evaporate and boil during testing) by first testing silicon oil (no evaporation) and then using sonde flow tubing that was made longer and narrower to create a restriction that matched the head pressure of actual solution in the cells.

 The paper finishes with showing (based on JMA long-term individual pump efficiency measurements) how sites using an average pump efficiency may have been calculating total column ozone too low by 2-4% around 2014 or specifically when the EnSci serial numbers within the 24000 series were used. 

Editing Suggestions/Comments/ and Questions:

Line 25:  Replace- “flew up” with “flown”

Line 27:  Replace- “chemical”  with  “electrochemical”

Line 27:  Replace- “The downlink of the data is taken care of by the radiosonde - also providing pressure, temperature, humidity and position measurements – the ozonesonde is coupled with.

with

“The downlink of the data, through the coupled radiosonde transmission, also provides pressure, temperature, humidity and position measurements.

Lines 33-34:   Reference – If available, please add a publication reference on the KI Carbon electrode type (KC) ozonesonde.

Line 37:  Replace- “ with take the ambient air into”  with    “ with bubble the ambient air into”

Line 39: Replace- “sampling air”  with  “sampled ambient air”

Line 39:  The flow rate is also needed to calculate concentration of ozone. Please add this in the last sentence to make it:  “The ozone concentration is calculated from this electric current and the volumetric flow rate of the piston pump.”

Line 46:  Replace “(4). Then again, the force of the piston takes the ambient air into the pump.”   with    “(4). The piston draws in a fresh sample of ambient air.”   

Line 46:  Replace  ”During the ozonesonde observation, this cycle is repeated”  with    “The cycle is repeated for each pump rotation. The steady pump speeds typically range from 2400-2600 rotations per minute (RPM).”

Lines 47-48:  I am having difficulty in understanding the first part of this sentence. I believe this is saying the back pressure is always the same from ground level to low pressure while ambient pressure is decreasing.

Line 63:  Replace “silicone membrane” with “bubble”   contraction

Line 67:  Replace “airbag contraction” with “airbag evacuation”

Line 67:  Remove this part of the sentence “and a gear pump with high pump efficiency”     The gear pump (nearly 100% pump efficiency) was only used at NOAA to confirm the accuracy of the  oil bubble flow meter.

Line 70:  Replace  “The system was designed to perform the entire series of measurement automatically, in order to be able to obtain pump efficiency with uniform quality.”   with    “The system was automated in order to  obtain pump efficiency measurements with uniform quality.”

Line 73:  Replace “ we could build up a”   with  “ we accrued a”

Line 104:  Replace “exhaust limit”  with  “minimum pressure”

Line  117:  Replace “Flowmeter controller”  with  “The flowmeter controller”

Line 118:  Replace “flow values of them” with “flow data”

Line 121:  Just a question on what is time-dense control?

Lines 130-133:  Figure 6 text:    Replace  “The bag is made of polyethylene in a volume of 140 ml.”  with  “The 140 ml. bag is made of polyethylene.”

Replace “in thermometers with  “by thermometers”

Replace  “measured in optical instrument”   with  “measured by an optical instrument”

Line 202:   replace  “back pressure”  with  “back pressure (load)”  .   This is a suggestion since back pressure and load are both used in the next sections.  I assume they refer to the same thing  so it would be good to include both in the title of section 3.3.

Lines 227-229:  Figure 8 text:  Suggest replacing “reaction solution”  with “silicon oil”  to be consistent with the text that notes silicon oil was used to represent the head pressure instead of actual sensor solution – which would create very large errors due to boiling of the KI/water solution.

Line 245:  Replace  “(sucked out)”  with   “pushed out”

Line 245:   Please add the typical pump temperature observed during a test.  For example, it would be helpful to know what the typical pump temperature at surface (beginning of test) and at the lowest pressure (3 hPa). 

Lines 335-336:  It appears that “reaction solution” is being used for referring to more than one thing.  It is used early in the paper when referring to the actual sensor solution (the KI salt water solution)  and then in line 335  it looks like in this text  “reaction solution” is referring to head pressure simulation of the sensor solution for NOAA/CMDL pump efficiency measurements, when NOAA/CMDL actually used non-evaporative oil to replace the reaction solution. Then in Line 336, I believe JMA is using extra tubing length to create a  simulated back pressure of the 3cc of reaction solution.  

It would be helpful to be clear where  “reaction solution” is actually back pressure or simulated head pressure of the 3cc of reaction solution.   

Figure 12:  Replace “UMYO 2002” with “UWYO 2002”  within the graph for Univ of Wyoming (blue line).

Figure 16:  The figure text letters (a) (b) and (c) should be in front of the data being described.  For example:  (a) Variation over time of pump flow rate.   

Review of Development of an automated pump efficiency measuring system for ozonesonde utilizing the airbag type flowmeter         

By Tatsumi Nakano and Takashi Morofuji

This is an important paper, and certainly appropriate for publication in ACP. I have just a few points that the authors should address before publication, primarily to do with clarity of explanation – there are a few areas where I was not entirely sure of the authors’ meaning.

Line 31: “…about 80% of stations of WMO...” In fact, all except Hohenpeisenberg and several stations in India use the ECC sonde.

Lines 60-61: I think it is important to make the point here that measuring the efficiency of each pump is NOT normal practice in ozonesonde launches, as it is considered difficult and time-consuming. As a result, almost all ozonesonde profiles are produced using average pump efficiency curves as described in the paragraph beginning in line 62. This is a source of uncertainty that the system described in this paper eliminates. It is a major advance and should be introduced here properly, as the scientific question that this paper addresses.

Line 65: Actually, it is the pump corrections that are underestimated, not the efficiencies.

Lines 139-140: Why does the airbag get wrinkles?

Line 157: I think you mean that the volume of the airbag when inflated is assumed to be the same regardless of ambient pressure, as long as the pressure inside is equal to that outside.

Lines 170-177: I find this description quite confusing. I see that there is some hysteresis, but it appears that the whole point of folding the inflation and deflation curves back on each other is to show that the inflation and deflation times are equal. Could this not be simply stated?

Lines 183-185: This is confusing, and the first sentence seems like it belongs somewhere else in the paper. I suggest writing simply: “The pump correction factor (the reciprocal of the pump efficiency) is obtained only from the time required for airbag inflation and deflation, and in the case of differential pressure ∆𝑝 is expressed from equation (2) as follows”.

Line 212: “the thin line”. Do you mean “the narrow tubing”?

Lines 220-223: I’m not sure that these remarks, or Figure 8c, add anything to the paper. Figure 8c is not mentioned further. The remarks are also confusing, coming in the middle of a discussion about “real-world” back pressures. I suggest dropping these lines, and Figure 8c.

Lines 255-256: Why is there a differential pressure? Is that because it is the method to determine when the bag is full/empty? It might be helpful to say this.

Lines 294-299: “…we found about half of the airbag temperature change rate affected the pump correction factor…”. So what happened to the other half? This paragraph appears to state that “our observations only followed Charles’ Law about halfway, so we used 0.5 as a fudge factor”. This is not acceptable.

Line 305: This contradicts the previous equation. Which one is correct? Is the pump change adiabatic or not? By the way, all equations should be numbered.

Line 306: Please explain what approximations were used to arrive at Equation 8. The reader should be able to reproduce your analysis without guessing.

Line 313: 𝑝𝑐𝑓3 (𝑝0) should be 1.

Line 330: Why use the measurements after #24000, rather than before #24000? You’ve just said that stability was not good after #24000.

Lines 351-352: I think you mean “for this experiment only”?

Figure 15 (upper) appears to be wrongly labelled on the x-axis.

Lines 403-406 (and Figure 17 caption): I am confused by this description. Should you not simply calculate the difference, for each sounding, in the total ozone found using your measured pump corrections to that using the average pump correction curve (either CMDL or your average before serial #24000 – or after serial #24000)?

Line 432-433: Can such an automated system be built and operated by other stations at a reasonable cost? Can it be commercialized? If so, this recommendation would carry much more weight.