The droplets vs aerosol debate for COVID19 is heating up. WHO has revised it's aerosol evidence review here:

https://www.who.int/news-room/commentaries/detail/transmission-of-sars-cov-2-implications-for-infection-prevention-precautions Section on airborne transmission.

I'm trying to understand what experiments would need to be run to

  1. determine if aerosol (long time, long distance) is possible.
  2. determine the "prevalance" of aerosol transmission. Here, let's say "<6ft" vs >6ft" infection.

Some snippets:

no studies have found viable virus in air samples.(29-36) Within samples where SARS-CoV-2 RNA was found, the quantity of RNA detected was in extremely low numbers in large volumes of air and one study that found SARS-CoV-2 RNA in air samples reported inability to identify viable virus.

[on superspreading events] However, the detailed investigations of these clusters suggest that droplet and fomite transmission could also explain human-to-human transmission within these clusters.

I'll provide my answer below, but I'll pick the best answer from someone else.

Why I ask

I'm trying to write some online risk tools and calculators. See http://tinyurl.com/covid3particles

search tags: covid19, covid-19, coronavirus, droplet, 5um

2 Answers 2


You can use Particle Image Velocimetry (PIV) (camera and laser sheet) to get a 2D image of velocity vector fields which show the velocity of plastic microbeads (whose size is known beforehand) to measure how long they remain suspended in the air. You can set up the experiment to measure the settling times for different sized microparticles. You cannot use (PIV) to measure the size of the particulates because the laser sheet used during the experiments has a Gaussian profile and finite thickness so you cannot easily provide a link between the intensity of light emitted by the particulates (and exposed to a camera) as they enter the sheet and the size of the particulate.

You can also use Phase Doppler Anemometry (PDA) to measure simultaneously the size and velocity of the droplets but this is a 1D point measurement system so you would need to apply the technique at several locations. Laser Doppler works under the same principle with the exception it only provides velocity measurements.

There are other more advanced techniques but these are the most common to measure velocity. Both techniques can be used to measure the background air flow conditions in both quiescent and turbulent flows, but there are technical and practical difficulties in setting up experiments which can measure both the continuous phase (air) and dispersed phase (particulates) simultaneously. If you don't introduce too many particulates in the air flow, then their presence does not adversely affect the background air flow (we call this "one way coupling") and therefore it would be reasonable to measure the air flow first and then measure the particulate velocities separately.

You can use basic imaging techniques with natural light rather than a laser to obtain estimates of the size of the particulates, but the light source would need to be powerful enough to illuminate the very small droplets or it would need to be very close to the source which may not make it practical.

You can find more information in common journals like Physics of Fluids, Journal of Fluid Mechanics and Experiments in Fluids. Numerous experiments have been carried out in fields such as combustion, drying, cloud physics etc.


A potential "ideal" experiment is a small, non-ventilated room. A known infected person is in that room for an hour or two. That person leaves and 5 minutes later a susceptible person (or persons) enter and stay in the room for several hours.

(This is not ethical, since it's not okay to knowingly make someone sick. It'd be nice to say CDC says it's mainly droplets and you won't get infected, but this won't pass ethics review.)

You'd also have to isolate that person for the incubation period (5-14 days, for example) to make sure they didn't accidentally get the virus from fomites or large droplets.

  • And that's the trouble... the ideal treatment+control experiment isn't ethical.

  • It's also hard to rule out "other explanations" if one uses observational data.

  • The "observation studies" of HVAC and cruises or Skagit choir have potential confounds (you can't prove they didn't get it from the cookies/snacks).


With regards to prevalence, I use the Jose Jimenez tool and the Wells-Riley model


This gives estimates of the quanta and a numerical probability of infection from aerosol. If there were a numerical probability of from droplet then the prevalance would simply be the ratio of the two.

  • Notice that only covid-19 counts for tags, "covid19" doesn't exists here in the tag system. You can also add coronavirus and any other available in the tag system. As moderator Carey said, this should be included in you question. Commented Aug 1, 2020 at 17:03
  • Editing. (thanks)
    – chongman
    Commented Aug 1, 2020 at 20:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.