Overall I recommend this manuscript for publication in this journal with minor corrections.

Please see attached file.

 Review of “VOLCANO3 - A Miniaturized Chemiluminescence Ozone Monitor for Drone-Based Measurements in Volcanic Plumes”

by Maja Rüth, Nicole Bobrowski, Ellen Bräutigam, Alexander Nies, Jonas Kuhn, Thorsten Hoffmann, Niklas Karbach, Bastien Geil, Ralph Kleinschek, Stefan Schmitt, Ulrich Platt

General Comments

This manuscript describes the development of an ethylene-chemiluminescence (ET-CL) ozone (O₃) instrument for use on small drones (UAS), for the purpose of obtaining measurements from volcanic plumes. The authors provide a technical description of their instrument and results from an initial field test at Mt. Etna. Application of CL techniques to measure O₃ in volcanic plume studies is an excellent idea since volcanic SO2 interferes with UV-absorption-based O3 instruments, and miniaturizing a CL instrument for drone use is novel.

Despite my enthusiasm for this endeavor, the manuscript, as written, contains major oversights that must be remedied to make it appropriate for publication. First, inadequate background on the chosen method, ethylene-chemiluminescence, is provided. This is important for several reasons, the most prominent being that “The method requires a constant supply of ethylene, which is a dangerous, flammable, and potentially explosive gas” (Long et al., 2014). What implications does this have for flying canisters of ethylene on drones? Were special permissions needed? How practical is use of this system? What alternatives might be possible? Second, the CL method has well-known interference from water vapor that is potentially significant for application to volcanic plumes (which usually contain abundant water vapor because water is the most abundant volcanic volatile species), but this is not mentioned. In addition to these significant oversights, there is little evidence presented that the system works well under ambient conditions (e.g. by comparison with a proven technique) and test data to indicate it works under plume conditions (e.g. high altitude, high humidity) is not provided. Finally, the results are poorly presented; even rudimentary comparisons to previous measurement campaigns at Mt. Etna are not considered.    

The manuscript needs significant revision to be appropriate for publication. Please see below for more detailed comments.

Specific Comments

A significant omission in the manuscript is that the well-known and dangerous nature of ethylene is never mentioned. This was a primary reason the U.S. EPA moved away from the technique, as described in Long et al., 2014 (pg. 5):

“3.1.3. Disadvantages The method requires a constant supply of ethylene, which is a dangerous, flammable, and potentially explosive gas typically stored in high-pressure gas cylinders. The use of such gas cylinders may be inconvenient and is often restricted by building fire codes or other monitoring site limitations.”

Also in Spicer et et al., 2010:

“The chemiluminescence method has been currently replaced in the United States by a Federal Equivalent Method (FEM), UV absorption (UV). A switch to the UV method occurred to reduce operational costs and improve safety by eliminating the flammable compressed ethylene gas required by the FRM.”

The hazardous nature of ethylene should be discussed and recommendations for safe handling. As described, is the setup subject to hazardous materials rules (e.g. UN1950)? What limitations might the use of ethylene present for the practical application of this method? Were extra approvals needed to comply with aviation rules to carry ethylene on a UAS? If special steps were taken for permissions to operate a UAS carrying hazardous material, it would be useful to describe them to understand the practicality and potential broad applicability of the method.

• 28: re-word the sentence “Besides its prominent role and abundance in the stratosphere, smaller amounts of O3 in the troposphere play an important role in the oxidation chemistry.” – perhaps “Besides its prominent role and abundance in the stratosphere, O₃ is an important oxidant in the troposphere.”

• 32: “…measurements of the vertical profile with high spatial and temporal resolution are rare, yet highly desirable.”

UAS offer an interesting new platform to potentially obtain O₃ profiles, but it would be appropriate to mention existing global ozonesonde networks here, such as those organized by GAW/WMO, NDACC, NASA, NOAA, and SHADOZ (many of these efforts and related publications are summarized at https://tropo.gsfc.nasa.gov/shadoz/index.html). For example, Stauffer et al. (2022) summarized 42,042 sonde profiles from 60 global stations that tracked O₃ from the surface to 30 km altitude from the years 2004-2021, demonstrating that considerable effort has gone into obtaining high-resolution, global O₃ profile data. Perhaps “rare” should be qualified, or the type of vertical profile that is meant could be clarified. From a practical standpoint it is important to note that most countries restrict the altitudes at which UAS can operate without special permission, which presents a significant limitation for using UAS to obtain vertical profiles. However, small UAS might have some advantages over other methods, such as tethered ballons, for low-altitude (i.e., boundary layer) studies.

• 35-37: “In fact, nitric oxide and ethylene CL measurements of O3 are still the standard method in the United States (USEPA, 2023) and are considered the most reliable O3 measurement methods (e.g. Long et al., 2014, Long et al 2021).”

This is partially true, but needs to be edited to clarify that the Ethylene-chemiluminescence (ET-CL) technique has been superseded by the Nitric Oxide-chemiluminescence (NO-CL) method, as summarized in Long, 2021:

“The ET-CL method is no longer used nor produced commercially and has been replaced by the NO-CL method…The ET-CL method was promulgated as the Federal Reference Method (FRM) for measuring O₃ in the atmosphere in 1971, and the NO-CL method was promulgated as the FRM in 2015 (U.S. EPA, 2015).”

Further information on the EPA’s rationale to move from ET-CL to NO-CL as a reference method is found on pages 65428-65429 in the U.S. EPA (2015):

“The existing O₃ FRM specifies a measurement principle based on quantitative measurement of chemiluminescence from the reaction of ambient O3 with ethylene (ET–CL). Ozone analyzers based on this FRM principle were once widely deployed in monitoring networks, but now they are no longer used for routine O₃ field monitoring… Although the existing O3 FRM is still a technically sound methodology, the lack of commercially available FRM O3 analyzers severely impedes the use of FRM analyzers…Therefore, the EPA proposed to establish a new FRM measurement technique for O₃ based on NO-chemiluminescence (NO–CL) methodology. This new chemiluminescence technique is very similar to the existing ET–CL methodology with respect to operating principle, so the EPA proposed to incorporate it into the existing O3 FRM as a variation of the existing ET–CL methodology, coupled with the same existing FRM calibration procedure.”

Also, I could not find “USEPA, 2023” in the references, nor could I independently find updated rules on O₃ FRM after 2015. The U.S. EPA webpage states “In December 2020, EPA decided to retain the current ozone standards set in 2015” (https://www.epa.gov/ground-level-ozone-pollution/setting-and-reviewing-standards-control-ozone-pollution).

The information about ET-CL being superseded by NO-CL is elided in the present manuscript, and more background concerning the chosen measurement technique (ET-CL) should be included. I suggest reviewing these references, fixing the USEPA, 2023 reference in the manuscript, and providing a more thorough background that explains why ET-CL was chosen for this application.

• 52-54: “The correction of the data with simultaneously measured SO2 (Kelly et al., 2013) or the application of selective SO2 scrubbers (Surl et al., 20l5; Vance et al., 20l0), however, are difficult and – at best - introduce significant additional uncertainty.”

I agree that using filters like those described in Surl et al., 20l5 and Vance et al., 20l0 for proximal plume measurements with high SO2 loadings is not ideal. However, it’s worth pointing out that both Vance et al. (2010) and Kelly et al. (2013) reported airborne measurements from dilute plumes where such interferences could be considered minor, and both studies found significant O₃ depletions in volcanic plumes that was much larger than any potential artifact:

In the case of Vance et al., several sets of measurements are included from a variety of techniques, but observations with the largest O₃ deficits come from airborne intercepts of the aged Eyjafjallajökull plume where co-measured SO₂ was less than 120 ppbv. This much SO₂ would result in a maximum of ~1-2 ppbv positive interference in a UV O₃ instrument which is negligible compared to 10’s of ppbv of O₃ loss relative to ambient levels (Vance et al., 2010, Supplement, Table 2).

Kelly et al., 2013 reported airborne intercepts of plumes from Redoubt Volcano, with SO₂ peak levels reaching up to only 1.2 ppmv, and most levels were lower. Their discussion points out the strengths and weaknesses and errors associated with their approach: see Section 3.4, e.g.

“Use of an interference-free technique to measure O3, such as chemiluminescence, would be preferable for making observations in volcanic plumes but unfortunately is not always practical. We acknowledge that the method we describe below has significant uncertainty but it has the advantage of using more common, inexpensive, and portable O3 and SO2 sensors to obtain information about O3 in SO2-rich volcanic plumes…”

However, Table 3 shows that the correction for SO₂ interference was generally quite small (a few ppbv O₃ – see O_3(raw) vs. O_3(correct)), and that in most cases even the uncorrected data showed lower in-plume O₃ levels than ambient O₃. In other words, the observed in-plume O₃ depletions in the presented measurements were generally larger than the artifacts introduced by SO₂ interference.  

Thus, diminishing these previous studies does not seem justified. Instead, it could be noted that such approaches are best carefully applied in dilute plumes, and that another approach (ie, chemiluminescence, like that described in the present study) is generally advantageous and could be considered necessary for measuring O₃ in dense and/or young volcanic plumes.

• 60: As written, it’s not clear what ‘traditional’ means here. This should be clarified with the expanded background that distinguishes between the CL-ET and CL-NO approaches.
• 82-83: Kern et al., 2020 should be added here, as they found very low BrO/SO2 ratios in the 2018 eruptive plume from Kilauea (see page 55), which corroborates with the low ozone depletion reported by Roberts (2018).

• 79-88: I’m not sure if I understand the model, as described, that predicts minimal ozone destruction in volcanic plumes (and the URL link in the reference did not work for me). Does this model assume constant influx of O3 into the plume? If so, it will not realistically capture the entrainment process during plume expansion. Furthermore, many “model studies with more evolved multiphase atmospheric chemistry mechanisms predict significant destruction of O3 in volcanic plumes”, for example many well-known works by von Glasow (e.g. 2003, 2009, 2010), Bobrowski et al., 2007, Roberts (2009, 2014), and more recent works by their collaborators. While O3 measurements in volcanic plumes remain rare - and many unknowns remain about in-plume halogen chemistry - most field measurements cited in the manuscript find significant O₃ destruction (e.g. Hobbs et al., 1982; Vance et al., 2010; Carn et al., 2011, Kelly et al., 2013, Surl et al., 2015, etc.). Thus, despite a paucity of in-plume O3 measurements, existing field studies tend to agree in general with published plume chemistry models that include multi-phase reactive halogen chemistry that is kick-started by mixing of hot halogen-rich volcanic gases with ambient air. The purpose of this section should be clarified, and adequate information and supporting references are needed if the intended purpose is to draw distinctions between models that predict or don’t predict ozone depletion in volcanic plumes.

• 100: “…with C2H4 being the most commonly employed reactant and which is used also in this study…”

Again, this is no longer the case and C2H4 has been superseded by NO (e.g. U.S. EPA 2015, Long et al., 2021). Please amend here and throughout. The choice to build an ET-CL instrument rather than an ET-NO instrument needs to be explained.

• “The principle of CL O3-Monitors” section: in addition to the theoretical description, this section should include practical information concerning the strengths and weaknesses of the technique. In addition to the problematic (flammable and explosive) nature of ethylene identified earlier, well-known positive interferences from water vapor in this type of instrument are not mentioned. This water vapor interference (which quenches the ET-O₃ reaction) is described in several references (e.g. U.S. EPA, 2015), including two already referenced by the authors (Kleindienst, 1993; Long et al., 2021). Example descriptions of the sense and magnitude of the water vapor interference are found in Kleindienst, 1993:

“The chemiluminescence-based monitors showed systematically higher readings than the UV monitors with added water vapor. The effect was found to be linear with water vapor concentration with an average positive deviation of 3.0 percent per percent H2O at 25 degrees C. For these measurement, ozone concentrations ranged from 85 to 320 ppbv and water concentrations from 1 to 3 percent (i.e., dew point temperatures from 9 to 24 degrees C). These results are largely in agreement with previous studies conducted to measure this interference, although the present study extends the range of water concentrations tested.“

                        A more recent summary is given by Spicer et al., 2010:

“Historically two methods have been widely used for ambient air O3 monitoring. The ethylene chemiluminescence Federal Reference Method (FRM) was dominant in the United States during the 1970s and 1980s. The only common documented interference to this method is water vapor. The extent of the positive bias is on the order of 3– 4% of the O3 reading for each percent (10,000 parts per million [ppm]) of water vapor in the air.”

Note: Spicer et al. (2010) independently found positive interference of 3 to 10 ppbv O3 per 10,000 ppmv H2O at O3 levels from 55-200 ppbv (3-9%) for a commercially available ethylene CL analyzer (Table 4).

Finally, the U.S. EPA commented on water vapor interference in ET-CL instruments when considering its rule change (U.S. EPA, 2015, pg. 65429):

“2. Comments on the FRM for O3

Comments that were received from the public on the proposed new O3 FRM technique are addressed in this section. Most commenters expressed general support for the proposed changes, although a few commenters expressed some concerns. The most significant issue discussed in comments was the relatively small but nevertheless potentially significant interference of water vapor observed in the ET–CL technique…However, in further response to these commenters’ concerns, the EPA has modified Table B–3 to extend this water vapor mixing requirement to newly designated ET–CL analyzers, as well. These measures should insure that potential water vapor interference is minimized in all newly designated FRM analyzers.”

Why is water vapor interference not included in the testing or error budget? In addition to describing the problem, simple solutions to the interference using simple commercially-available dryers are tested and analyzed (e.g. Long et al., 2021). According to a text search, the only mention of “water” in the manuscript is l. 70, where it is listed as the first (and presumably most abundant) component of primary volcanic gas emissions. This is potentially very important given how water-rich volcanic plumes are. Plumes routinely contain 1000’s to 10,000’s ppmv of water vapor (especially close to the source), which suggests that the reported ET-CL measurements could have potentially significant positive artifacts, unless this issue was mitigated somehow. Please describe any testing or mitigation tactics for dealing with water vapor during the development and field testing, how artifacts are dealt with in the results, and please clarify if water vapor was measured as part of the sensing package.

• 124: Please include a photograph of the instrument to accompany the schematic shown in Fig. 1. For example, Figure 13 from Bräutigam, 2022 might be appropriate.
• 153: If I understand Figure 2 correctly, the dark current appears to vary from ~4-12 ppbv equivalent O3 at temperatures from 15-35*C. How is the derived uncertainty of the fit so small (“1 ppb”)?
• 171: “To calibrate the monitor several calibration measurements with varying O3 mixing ratios in different sequences are made.” – this is too vague. Please elaborate on the calibration conditions and procedures. Also indicate if the calibration procedure utilized dried or humid air.

• 175: Change to “The detection limit…”

• 180: here and throughout, make sure to include all equation variables and units, and be consistent (e.g. ”ppm” appears 9 times in the manuscript, “ppmv” twice; “ppb” 20 times, ppbv once).

• 200: Was anything done to test and validate the instrument’s performance at different elevations? Was it compared to a reference instrument? If so, please include and explain how this was done. What was done to assure the instrument would work well in plume conditions (ie, high elevation, humid)?

• 200: Was a vertical profile flown to evaluate atmospheric structure at altitudes relevant to the plume?

• 221: please include more information on the “little-RAVEN”. What sensors were included? What were the ranges/resolution/etc. What was the weight? Was it flown simultaneously with the VOLCANO₃?

• 230: Was the atmosphere stratified or compositionally heterogeneous with respect to the point of volcanic gas emission vs. the point of measurement in the buoyant plume? What was the difference in elevation from the vent to the altitude of the plume measurements? The manuscript only considers chemical ozone destruction and does not consider the impacts of mixing, entrainment, and transport of air from chemically dissimilar air parcels on measured O3 (as described by Kelly et al. 2013).
  • Were any BrO measurements made coincident with the O3 measurements? These would help to link O3 depletion to BrO formation.  

• 230-253: the results section is weak. Ozone depletion has been measured many times (relatively speaking) at Mt. Etna. How do these new results compare to previous measurements and models (e.g. Vance et al., 2010, Roberts et al., 2014, Surl et al., 2015)? Did the CL technique obtain different results than these previous studies?

• 249: The SO2 sensor range was only 16 ppm? Why? This seems like a serious limitation of the setup.

• 255: Improvements:

As noted above, C2H4 is flammable and problematic from a hazardous materials standpoint. Is cyclohexane (C6H12) also problematic from a safety/hazardous materials standpoint? What about other methods such as NO-CL or so-called ‘scrubberless’ methods where N2O is converted to NO and used to titrate O3 (described in Long et al., 2021)? Also, what would need to be done differently next time to better characterize the plume and its chemistry?

• L294: “Today, our knowledge is mainly based on model studies. Such model studies show for instance a complete ozone depletion in the centre of halogen rich volcanic plume after a relatively short distance from the emission point (about 10 min downwind, Roberts et al., 2014) but a solid experimental prove of those theoretical consideration is still missing.”

I disagree with this conclusion and suggest revisiting Roberts et al., 2014. They present several scenarios, some of which result in modest ozone depletion (see their figure 8). It’s true that only one study (thus far that I’m aware of) has coupled airborne measurements with model results (Kelly et al., 2013). In that case the model parameterization was constrained as best as possible by field measurements and did not predict complete ozone destruction in the plume. In fact, good overall measurement-model agreement was found, suggesting that the model captured the major pieces of the chemistry in that case. Unfortunately many of the most important species involved in the relevant chemistry are hard to measure, so models are critical for understanding these unique systematics. This comment should be reconsidered, although I agree that more measurements are needed.

L.303: Given the flammable and explosive nature of ethylene, is it reasonable to suggest flying such a payload over fires or cities? The conclusions need to reflect the limitations of the current approach.

Table 1:

• Can a plume age be calculated for the listed measurements?
• Why is the correlation coefficient and r² value listed? Isn’t this redundant?
• How do the derived O3/SO2 ratios compare to other measurements from Etna (e.g. Surl et al. 2015)?
• What was the humidity inside the plume?
• At which altitudes was the plume measured?

Figure 2:

• What are ‘normal’ measurements? Please clarify or refer the reader to the text where this is described.

REFERENCES

Kern, C., Lerner, A.H., Elias, T., Nadeau, P.A., Holland, L., Kelly, P.J., Werner, C.A., Clor, L.E.,

Cappos, M., 2020. Quantifying gas emissions associated with the 2018 rift eruption

of Kīlauea Volcano using ground-based DOAS measurements. Bull. Volcanol. 82,

1. https://doi.org/10.1007/s00445-020-01390-8

Long, R., Hall, E.S., Beaver, M., Duvall, R.M., Kaushik, S., 2014. Performance of the

Proposed New Federal Reference Method for Measuring Ozone Concentrations in

Ambient Air Technical Report, U.S. Environmental Protection Agency, EPA/600/R-14/432

Long, R. W., Whitehill, A., Habel, A., Urbanski, S., Halliday, H., Colón, M., Kaushik, S., and

Landis, M. S.: Comparison of ozone measurement methods in biomass burning smoke: an evaluation under field and laboratory conditions, Atmos.  Meas. Tech., 14, 1783–1800, https://doi.org/10.5194/amt-14-1783-2021, 2021

Stauffer, R. M., A. M. Thompson, D. Kollonige, D. Tarasick, R. Van Malderen, H. G.J. Smit, H.

Vömel, G. Morris, B. J. Johnson, P. Cullis, R. Stübi, J. Davies, and M. M. Yan (2022). An Examination of the Recent Stability of Ozonesonde Global Network Data, Earth and Space Science, 9(10), https://doi.org/10.1029/2022EA002459.

Spicer, C.W., Darrell, D.W., Ollison, W.M., 2010. A Re-Examination of Ambient Air Ozone

Monitor Interferences. Air & Waste Manage. Assoc. 60:1353–1364DOI:10.3155/1047-3289.60.11.1353

U.S. Environmental Protection Agency (EPA): National Ambient Air Quality Standards for

Ozone, Federal Register, 80, available at: https://www.govinfo.gov/content/pkg/FR-2015-10-26/pdf/2015-26594.pdf, 2015. 

This manuscript describes the development of an O3 monitor using ethylene chemiluminescence and its application overcoming challenges measuring O3 in volcanic plumes, highlighting the lack of SO2 interference versus more common UV measurements.

Both the development of the instrument and uses shown in the manuscript are interesting and in scope for AMT, but I feel that there are areas that require substantially more discussion before final publication.

Major Comments:



- There is no discussion of the effects of water vapour on the system, despite the listing of water vapour as the primary gas emission from volcanoes.  This needs to be addressed, as there are % level biases reported for ethylene-CL with water vapour

- Section 3.2.1 describes how the PMT’s dark current is corrected for based on a fit with cell temperature rather than maintaining a constant temperature. L150 states that the fit parameters “are determined at regular intervals” which is very vague. It is not clear to me if this is something that has been determined in the laboratory, or whether the instrument is capable of determining this during normal operation. If this is the latter, I don’t see from the instrument description or figure 1 how this would be achieved.

                - Figure 3 shows the measured dark current, and within the relevant temperature range there is variation of +/- 5 ppbv of O3 from the fit. It needs to be more clearly demonstrated how the authors arrive at only 1 ppb of additional uncertainty from this.

                - More broadly a better diagram of the instrument would substantially improve the readers understanding. For example, the authors note that the ethylene flow is controlled via capillary, but do not state how flow through the instrument is controlled or measured – though this is required for their correction factor Ccon. Without sufficient control would this not vary with altitude?

- No mention of the time resolution (data acquisition time or averaging intervals) of the instrument is mentioned. From figure 6 data is presented at what appears to be ~2 Hz? This is relevant broadly for the reader to understand the instruments capabilities, but also for the context of performance statistics such as LOD. LODs reduce with averaging, and the reported LOD of 1.13 ppb appears low for a 2 Hz measurement, with the uncertainty from dark current alone presented as 1 ppb. Moreover the stated LOD assumes the standard deviation in the dark current to be 0.4 mV, which from figure 3 appears to be a favourable case? Do the authors have data that they could use to perform Allan variance to provide further information on the LOD?

- Section 3.2.4 Instrument response time describes the experiment as “swiftly connecting the hose to the monitor”. More details are required on how this was conducted – was the process automated using a fast acting valve? If this was performed manually, I would expect this response time to be biased high.  A good understanding of the time response is relevant to the previous point surrounding data acquisition and averaging, and should be put into context with how those data are presented.

Technical Comments:

- Reference need to be checked, for example: L36 USEPA 2023 does not appear in the reference list, the DOI for Kleindienst 1993 returns “not found”, the 2B Technologies manual is not referenced in the text.

- Figures in the SI are referenced out of order to that in which they are presented  e.g S4 is referenced on L88, S1 is not referenced until L234

- Data availability – I see no reason why this data cannot be archived (along with processing code?) and referenced in this section, as per the AMT data policy?

- L37  “nevertheless, nowadays and since many decades” needs rewording

- L127 includes text from the figure caption for Figure 1.

- L128 / 130 / 131 – the labels that this text is referring to do not appear on the figure, and instead seem to refer to Figure 13 of Bräutigam 2022.

- L128 t(h)rough

- L 211 “0.6 million years”  should this read “0.6 million years ago”?

- L233 “Fig Figure 5”

- L266 The chemspider reference does not appear in the reference list, and I suspect provides a reference to the data where this is determined?