A Drone-Based Bioaerosol Sampling System to Monitor Ice Nucleation Particles in the Lower Atmosphere

Round 1

Reviewer 1 Report

General comments:

Terrestrial ecosystems emit a large number of bio-aerosols, an important source of IN to influence cloud and precipitation formation, and gaseous precursors (e.g. VOCs) to transverse into SOA by chemical processes to act as CCN or IN. These biological aerosols have pollen grains, fungal spores and bacteria cells, with a relatively larger size compared to other type aerosols. Now the IN formation and its influence on cloud and precipitation even climate is still an open question. The paper presents a self-developed system to measure IN and its transport near sources, using quadcopters Drone-base Aerosol Particles Sampling Impinger/impactor. The paper uses the instrument to observe INs in Austria, and attempts to ensure validation, sampling efficiency and field data. The results enrich studies on measurement technology of biological INs in the world. The topic of this paper is of common interest within the scientific community. Although the manuscript includes some important data, however, the quality is not sufficient in the current state to be directly published. The authors should take the suggestions made here into consideration for revision.

 

Specific suggestions:

The abstract is too simple, however, a long content is on INs knowledge introduction. In lines 305-308, Figure 6b shows the cumulative nuclei concentrations summed up over 3 measurements…, but there are only two measurements seen. The instrument still needs more campaigns to validate its ability, such as cases of higher INs concentrations, larger height (more than 50m), etc. The authors should explain it.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

This paper presents a methodology for sampling and studying biological Ice Nucleation Particles (INP) using a small rotary wing drons. The authors describe the methodology quite accurately and demonstrate the performance using field observations in a sampling site located in a mountain area of Austria. The paper is clearly presented and overall quality of the manuscript is very good. Some proofreading for typos is however needed. I also have a few main remarks. First, although the sampling altitudes are quite low, it will be useful to know if the sampling flow-rate is affected by changing pressure. As far as I understand, the impactor efficiency is validated only at the ground pressure. But, for example, what can be effect of changing pressure on the cascade impactor sampling efficiency (particles size selection at each stage)? Second, more quantitative comparison of the obtained values with those known in the literature and putting the results in context of other studies could strengthen the paper. For example, the results of freezing experiments can be discussed and compared with literature in more quantitate way, avoiding that the readers need to search in the provided references. Last, a discussion can be welcome on how this local and low altitude sampled particles represent properties of ice nucleation particles after transport and aging processes.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

General Comments:

The paper is understandable. This is a very important work; however, difficult area of research. This paper encompasses a lot of work and effort. Understanding how much work has gone into this paper, I still need to recommend rejection of the paper. I feel that the conclusions are not supported by the methodology used and data presented. Many items in the paper point to the opposite conclusion in my opinion that what is stated.

For example, the first sentence in the conclusion, “We conclude DAPSI to be an efficient tool to sample bioaerosols.” My take is the it is “not” an efficient tool.

A recommendation is to clearly state a hypothesis at the start of the paper. It is great that the paper states a “goal”, which is “The overall goal of our work was to develop a low-cost, lightweight system to sample bioaerosols”. This has clearly been done; however, it really impossible to not do this in some way. Better to state a hypothesis that could be false and be willing to accept the result of either way. A possible hypothesis statement is: “It is possible to develop a low-cost, lightweight system that is able to sample bioaerosols as effectively as a state-of-the-art commercially available sampler.”

A real issue is with comparing the developed system, which is inexpensive to costly commercially available system. My feeling is that the “low cost” system is not close to the performance of costly commercial system. While to be fair, the paper does not present such a comparison. I would suggest redoing the paper and just focusing on what has been done to develop the system, which is a lot, but there is still lots more to be done to be an “efficient tool”.

The methodology (PSL sphere) for determining the sampling efficiency is insufficient and is not the standard method for determine efficiency. The method presented for determine efficiency is the difference between the particle size distribution with and without the impinger. This method does not allow determination of where the particles went, and if the particles really ended up in the collected sample. For example, the impinger is made of plastic that is not conductive; hence, the wall of the tubing/impinger can collect and hold charge particles. When is the water used? Is water placedI still need to recommend rejection of the paper. I feel that the conclusions are not supported by the methodology used and data presented. Many items in the paper point to the opposite conclusion in my opinion that what is stated. in after sampling to collect the particles or before sampling starts. There is no reason to believe that all the particles will end up in the water sample. What needs to be done, is to do the ratio of the mass of particles that enters the impinger to the mass collected in the sample. Measure the mass using well know method and compare to impinger sample. The “sampling biological INP” seems to be the way to get efficiency (section 3.1.2), hence, it is the size efficiency that I have issues with. For section 3.1.2, it hard to understand if the mass used in this test is similar to the mass that would be sampled in the atmosphere in ten minutes. Would not the mass (or number of ice nuclei) of biological particles have to be the same in the efficiency test as in the atmospheric sampling?

What is the point of using an impinger instead of a filter system or the cascade impactor? The ice nuclei measurements should be able to be done on filter samples.

What is the measurement objective? Is it to determine the amount of mass of biological particles. The number concentration of ice nuclei? The type of biological particles? The source of the biological particles?

In some work I’ve been involved with, we been doing DNA sequencing of atmospheric particles collected on filters. DNA sequencing is now relatively inexpensive, only costing $100-$200 per filter, and you can sequence unculturable bacteria and fungi. Knowing the species, allows measurements to be related to the source. Hence, I recommend considering using DNA sequencing to determine source of sampled particles.

The paper looks at fluoresence properties and ice nucleation activity; however, I am having a hard time seeing the link of these measurement to science questions. Is the idea to compare drone measurement over land to that over water to determine differences of ice nucleation activity? There can be a lot of mixing in the atmosphere and the biological particles can last for hours in the atmosphere so would not expect a large difference with location.

One major advantage to low-cost science instrument is for many people to be able to use them. Use of most science instruments today require software, which could sufficiently affect scientific conclusions. Hence, I feel that all scientific software need to be in open software repositories. Putting software in software is free (for example GitHub or SourceForge repositories) and easy to do (less than 1 hour of work). I don’t believe papers like this should be published if the software is not released in a public repository. As a review, I don’t want to see your Flowchart, I want to see your source code and check the code. We need to do more to check scientific software.

Lot of people don’t do it; however, my opinion is that papers are much easier to understand if what is done for the research presented in the paper is written in present tense and not past. Use past tense for previously published work.

 

In summary, I’m recommending rejection because I don’t believe the conclusions follow from the presented measurements and methodology used. However, the idea of the project is good, and the work conducted is useful for others to know so they can learn from this project. I would suggest revising the conclusions to reflect what is presented in the paper and include a clear, testable hypothesis at the state (not just a goal). Be willing to accept a negative result of not being able to prove the hypothesis. While people like positive (we can do this) results, I like to see more things that didn’t work as expect published so we don’t waste time trying what has not worked and use our time improving things. There is lots of good stuff here, but things need to be improved upon with additional effort. Great that this system was used to do some sampling; however, I don’t see what science it was able to do (what hypothesis it was able to address). State this and say there is more work that needs to be done.

Additionally, I really have an issue with method used in Section 3.1.1. I don’t see knowing the efficiency of different size particles as that important. Yes, it possible to do, but would require lots more work that has been presented. I would remove 3.1.1 and focus on more description/analysis of 3.1.2. While I am recommending rejection; however, I don’t believe it would be too much work to write an acceptable paper, and would suggest doing so. Hope my comments are useful, some detailed comments are given below. I’ll sign the review so happy to discuss further sometime, maybe at EGU conference.

 

Specific Comments:

What is Imp? Not defined in text. Assume it means Impinger (Imp) from figure 2. Why even use Imp as an acronym, it is not common used term. All figure caption should define all acronyms in the caption since the Figure should be independent. People read figure captions independent of the text and hence all acronyms need to be define in each figure caption, even if the acronym is defined in text.

Figure 1: Pictures are o.k.; however, better to have a diagram of the components, and details of each part. For example, where is the valves in the sampling system. What are the material and size of components?

I know people use it but I don’t understand the term “aerosol particles” as in “primary biological aerosol particles”. An aerosol is a suspended particle in the atmosphere. So say “aerosol particles” is “suspended particle particles”. What do the author mean by including “primary” here. What would a secondary biological aerosol be? We use primary and secondary particle to make a difference between particle emitted to the atmosphere and particle formed in the atmosphere. How can a biological aerosol “form” in the atmosphere. Such biological aerosols can break into smaller aerosols but how can they form in the atmosphere?

Section 2.1.1: Is not the “polyvinyl chloride (PVC) tube going to remove particles. Is this same setup used for the efficiency tests? Using PVC tubing, particles will collected on the walls of the tubing since it is not conductive. Why not use conductive flexible tubing?

Section 2.3.1: The TSI 3076 Constant Output Atomizer is not good for generating PSL 2 um particles. The 3076 will break the particles and you get a wide distribution, with lots of small particles. Use a aeromist nebulizer, make sure it is clean and use ultra pure water. Best to use new nebulizer, since they are hard to get clean.

Section 2.1.1: “An aliquot of 15 ml of ultrapure water was used as the collection buffer ..” Explain further how this was done. Was the 15 ml used to remove particle for the walls of the plastic vials? Was the 15 ml just placed in the bottom of the vial? Is it not assumed that all the particle collected ended up in the 15 ml of water?

Line 141: How was the OPC connected into the sampling system? Was an OPC on each drone? Please include in diagram.

Line 141-142: Was the flow sensor used to determine the volume of air sampled? Was the flow sensor calibrated

using a traceable standard?

Figure 5: This spectrum (a) shows why you can’t use the TSI 3076 to generate particles. Use a nebulizer and measure the spectrum, you’ll get a much better peak at 2 um.

Figure 6: Define all acronym in caption. BP-WW) is not defined on first usage in the paper.

Line 204: “… with two lmp vials in a row”. What was the length and type of tubing between the two vials. Conductive tubing will remove particles; hence, the efficiency you get with this method will depend on the type and length of tubing. The tubing will remove lot of particles. What is the efficiency determined with this method, using a very short tubing (like 10 cm) compared to long tubing (10 m)? The loses in the connecting tubing is one reason this method can’t be used for determine efficiency.

Table 1: Is the height above sea level or above ground? What is CEST? Define in caption. Use complete sentence in caption, for example “Flight Data of the field campaign in Upper Austria with DAPSI has no verb.

Line 298: “Both samples of the Imp vials...” Where are these two Imp vials from? What setup was used to conduct measurements? Were they sampling the generated BP? What was the concentration of particles and flow rate used in this measurements. Was the “Water” sample put in a clean (with no air, or filter air sampled) vial before doing the ice nuclei test?

Line 319: GPS altitude is much better that pressure altitude, there is no GPS measurement available on the drone? GPS measurements are not expensive. For example, InterMet_iMet-XQ2 give GPS, temperature, humidity for $500.

Line 320: So Po is ground altitude so h is above ground level?

Table 3: State in caption, what instruments made each measurement.

Figure 7/8: Use describe names for the zones, instead of 1, 2, 3. For example, Lake Zone instead of Zone 1.

Line 368-369: “A picture of a control aluminum foil, which was not exposed to air” Can you provide more information about the “control”. I would think that a control would be exactly the same as a sample, expect without the valves being open. I would suggest that you have to do the flight and everything else the same. Is your control just for the instrument and not the sampling methodology?

Line 371: Cloud not the “small sized bioaerosol” be broken large particles? Will not large super micrometer sized bacteria break when sampled by the impinger?

Figure 10: What does UA1 and IMP1 mean? Assume IMP1, IMP2 and IMP3 are for the different zones. To me, there is no difference between zone 1 and 2 as I assume the authors would expect. Hence, I would conduce that the sampling methodology is “not” effective for seeing the difference between these location and differences in sources.

Line 435-444: Seem to indicate to me that these small drones are not able to collect enough material and hence can’t provide the scientifically required measurements.

Line 472: “DAPSI to be an efficient tool to sample bioaerosols.” How can DAPSI be “efficient” if it does not provide measurements that can address a scientific question/hypothesis? What exact hypothesis can DAPSI be used to address?

Line 478-482 – Don’t understand this living bioaerosol application. Why does the current system exclude living bacteria?

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

General Comments:

There were a lot of improvements in the paper. Thanks for the explanation of many items. The current paper’s focuses on presenting a method for using unmanned aircraft to sample bioaerosols. This scope is good and useful material to have presented in a paper. The items that resulted in my rejection recommendation have been removed/adjusted. There are still items that could improve the understanding of the paper; however, nothing at level of recommending rejection. Hence, my recommendation is accept with minor revision. I suggest that the authors review the below comments that I’m still confused about and consider how things might be improve in this paper or future work. These are minor issues since they makes things more difficult to understand but of do not lead to incorrect conclusions. I know many people do them, and they will continue to do them because they make things difficult to understand but won’t result in paper rejections. Hence, many reviews just don’t have the time to comment on them. First round comments in black text, author’s comments in green text, new comments in ‘red’ text.

Referee #3: The methodology (PSL sphere) for determining the sampling efficiency is insufficient and is not the standard method for determine efficiency. The method presented for determine efficiency is the difference between the particle size distribution with and without the impinger. This method does not allow determination of where the particles went, and if the particles really ended up in the collected sample. For example, the impinger is made of plastic that is not conductive; hence, the wall of the tubing/impinger can collect and hold charge particles. When is the water used? Is water placedI still need to recommend rejection of the paper. I feel that the conclusions are not supported by the methodology used and data presented. Many items in the paper point to the opposite conclusion in my opinion that what is stated. in after sampling to collect the particles or before sampling starts. There is no reason to believe that all the particles will end up in the water sample. What needs to be done, is to do the ratio of the mass of particles that enters the impinger to the mass collected in the sample. Measure the mass using well know method and compare to impinger sample. The “sampling biological INP” seems to be the way to get efficiency (section 3.1.2), hence, it is the size efficiency that I have issues with. For section 3.1.2, it hard to understand if the mass used in this test is similar to the mass that would be sampled in the atmosphere in ten minutes. Would not the mass (or number of ice nuclei) of biological particles have to be the same in the efficiency test as in the atmospheric sampling?

 

Answer: The same methodology to determine sampling efficiency has been used in previous studies for nearly identical impinging systems [Powers, C. W., et al. (2018). "Coordinated sampling of microorganisms over freshwater and saltwater environments using an unmanned surface vehicle (USV) and a small unmanned aircraft system (sUAS)." Frontiers in microbiology 9.] This technique gave very promising results for such impinging systems and therefore has also been applied in this study. The most precise way is to determine particle concentration before and after the sampling system as we did in section 3.1.1, while analysis of water suspended particles is not practicable. Due to our results, we assume the loss in the tubing negligible compared to the sampling technique. We have tried out the measurements of the particles with different lengths of tubings and did not observe changes in the concentrations.

A critical issue is the plastic used for the Impinger setup. However a conductive material as proposed by the referee was not applicable, since the used plastic vials are 1) cheap, 2) sterile, 3) lightweight and 4) easy to change during field measurements. Moreover, the sampled air is just in contact with the glass surface of the pipette before it is lead through the sampling medium. Since we do not state quantitative numbers of INPs, we do not want to change this part of the paper.

The mass used in section 3.1.2 cannot be directly linked to INPs aerosols present in our atmosphere since the concentration and size of INPs can vary over several orders of magnitude depending on the different sampling spots. However, to do a precise characterization of DAPSI and to validate the system for field measurements with quantitative output, the best way would be to measure during field campaigns with standard equipment. We will do so in upcoming studies, but it goes beyond the scope of this publication.

 

I still am concern about particle lost. Great that different length of tubing did not affect the results. The efficiency discussed in paper is a “removal” efficiency, not a “collection” efficiency. These are not necessary the same thing. I understand that it is difficult to determine a “collection” for INP. I would suggest looking into this, if at all possible, in future work. Maybe some lab comparison between Impinger and Impactor on controlled concentrations of bioaerosols.

 

Referee #3: One major advantage to low-cost science instrument is for many people to be able to use them. Use of most science instruments today require software, which could sufficiently affect scientific conclusions. Hence, I feel that all scientific software need to be in open software repositories. Putting software in software is free (for example GitHub or SourceForge repositories) and easy to do (less than 1 hour of work). I don’t believe papers like this should be published if the software is not released in a public repository. As a review, I don’t want to see your Flowchart, I want to see your source code and check the code. We need to do more to check scientific software.

 

Answer: We wanted to achieve, that the reader is able to understand how the software works in order to re-build DAPSI or to build similar sampling systems. A flowchart provides a general overview, also for readers which are not specialists in programming or write in other programming languages. Source codes are often confusing for programmers that use different languages or programming styles. However, if someone wants the source code we will provide it on a public database.

 

Great if the paper could indicate that software is available upon request. This is typically what has been done with data set, first people would indicate the data set used is available upon request, and now that is not “good” enough in many cases and the data set need to be in an open repository. Should we reject papers that don’t make software openly available that is at the heart of an science hypothesis presented in a paper? Many of us understand code in whatever language you write. Also, comments and coding style should make it easy to other to understand the code.

 

So taking a close look at the flowchart presented. What is the meaning of the “Imp. On?” check since if it is on (Yes) then it just get set on, if it off (No) then it get set to 0 (Off). Thing changes with this check as I understand it. The “Flow Switch is to determine what measurements are stored. Why not save all the data all the time? I can’t image that the USB sticks are not large enough. I save voltage measurements from an analog-to-digital converter even when things is connected to it in my software, just in case something is connect to it at a later time. Without the available source code, I can’t check that the flow diagram is right. Having the diagram presented, only indicates that not all data is corrected at all times, which is something I would never do. I don’t think many readers are interested in such a diagram. I’m not sure how it would be at all useful if I was trying to build a similar system or understand a program. We don’t present/review flow diagrams in our University programming classes or class assignments. I look at the student’s code, test student code.

 

Hence, my opinion is Figure 3 is not needed.

Referee #3:I know people use it but I don’t understand the term “aerosol particles” as in “primary biological aerosol particles”. An aerosol is a suspended particle in the atmosphere. So say “aerosol particles” is “suspended particle particles”. What do the author mean by including “primary” here. What would a secondary biological aerosol be? We use primary and secondary particle to make a difference between particle emitted to the atmosphere and particle formed in the atmosphere. How can a biological aerosol “form” in the atmosphere. Such biological aerosols can break into smaller aerosols but how can they form in the atmosphere?

Answer: This term is indeed often used in literature since it makes sense to highlight that particles are meant. The term “sampling aerosols” would include the gas phase, which was not sampled. So “aerosol particles” is the most common and best suited description

I strongly disagree with the author’s statement that “aerosol particles” is best suited. The author just state this is “often” used, please provide a source for this definition. I believe it is “often” used incorrectly. The American Meteorological Society Glossary, see http://glossary.ametsoc.org/wiki/Aerosol, has a definition that does not follow that author’s use of the term. Furthermore, I’ve survey 10-20 people in atmosphere science, meteorology, and chemistry. Done of the scientist I’ve asked would consider an “aerosol” to ‘include the gas phase’. I don’t see how an “aerosol” could include gas phase material, is it the material associated with solid phase? Just the concentration in the gas phase? My understanding is that all solid material have a “little” material in the gas phase so you would have to measure this very small concentration in the gas phase to be measuring an aerosol?

Hence, I an confused as to why the authors believe an “aerosol” include gas phase material? The ‘author’ answer would be considered incorrect in all the meteorological classes we teach, both at the undergraduate and graduate level. I am not sure where this usage comes from or why. If it because of different fields using different terms; however, a large amount of the readership of the paper will be confused by the use of “aerosol particle” as I am. I have not reviewed the many other papers that have used this term, only this one, so just because other papers have done it “wrong” does not mean this one should. We either need to stop using the term incorrectly or have a definition that scientist learn. Why not just use “particle”? What is the “aerosol” meant to convey that particle does not? The simple definition of an “aerosol” is “a suspended particle in the atmosphere”. With this definition, an “aerosol particle” would be a “a suspended particle in the atmosphere particle”. Aerosol include the idea of a “particle” to me and scientist I’ve talked with. Provide a reference or more information of authors usage/definition of the term “aerosol particle”.

Primary means, that these particles are emitted directly from biological processes over land-surface. Secondary organic aerosol includes aerosol particles which are formed or transformed in the atmosphere during chemical and/or physical reaction. Also this term is commonly used in literature.

I understand primary and secondary for organic aerosol; however, why apply the term to bacteria and fungi (bioaerosol) since I can not understand how a bacteria could be “secondary”? If you have primary bioaerosols, there would have to be secondary bioaerosols, which I do not understand. I am trying to understand what the authors mean with the term “primary bioaeosol” that is not covered by “bioaerosol”. Why only over land surface and not water, for example bubble breaking in rivers to release bioaerosols. What would a “secondary” bioaerosol be?

Author Response

We are very happy about the positive feedback of referee #3.

We have answered all the questions and comments in detail (please see our uploaded answers).

The main amendments are:

a) Referee #3 agrees to shift the problems regarding the sampling efficiency into a forthcoming paper and we have changed the text respectively.

b) The flow chart of the software program has been shifted to the appendix and the source code will be available on request to the corresponding author.

c) The term "aerosol particles" has been defined now more clearly within the paper.

 

We are very thankful for your patient and we are looking forward to the publication of our manuscript in Remote Sensing.

On behalf of all co-authors,
Hinrich Grothe

 

Please see more details response in the attachment.

Author Response File: Author Response.docx