Post #829: COVID-19 cases, weather, and heated outdoor restaurant seating.

Posted on September 28, 2020

Source:  Calculations from NOAA temperature data for Reagan National Airport, COVID-19 case counts from the Virginia Department of Health.

This post take the place of my normal posting on trends in new COVID-19 cases in Virginia.  It really an opportunity to gather together a few bits of information about heat, humidity, and transmission rates of COVID-19.

The quick upshot is that the environment that maximizes your risk of transmission of COVID-19 is cold, dry, stagnant air.  In short, for avoiding COVID-19, the warmer, the wetter, and the windier the better.

But what about using heaters to extend the outdoor dining season?  Electric heaters will reduce the relative humidity of the air.  Doesn’t that raise the risk of COVID-19 transmission. 

My answer is no.  I’m almost sure that extending the outdoor dining season with heaters provides a net benefit in suppressing spread of COVID-19.  The “almost” there isn’t due to the behavior of the virus, but to the behavior of diners.

Details follow.  If your sole interest is in outdoor restaurant dining, skip to the end.

A slight downward trend in daily new COVID-19 cases in Virginia.

Below are two of my standard charts for Virginia.  The first shows Virginia (blue) and Fairfax County (orange).  A downward trend in new cases per day is visible to the eye.

The second contrast the late reopening areas (blue, NoVA + Richmond), against the rest of the state (orange).  Again, slight downward trend in both areas, visible to the eye.

But don’t you have to ask yourself, why?  Why, after months of stability, are we now seeing a gradual decline in cases.

The reverse summer effect, and a little bit of lying with statistics.

It did not go unnoticed that this summer, the worst COVID-19 outbreaks were in areas where you more-or-less have to live inside, in the air conditioning, due to the heat and humidity.  This was dubbed the “reverse summer effect”, as reported in the Washington Post.

This “reverse summer effect” has nothing to do directly with the heat and humidity.  It has to do with outdoor settings and open windows being a lot safer than indoor settings and stuffy rooms, in terms of spread of COVID-19.  You can see a brief summary of the evidence on that, from China and Japan, in this article in The Atlantic.  (And that’s almost certainly due to aerosol spread of disease, despite the fact that the CDC can’t manage to say the A-word in public.)

But whatever effect there is, it will have to be a Goldilocks effect.  Too hot, or too cold, drives people indoors.  Only when the temperature is just right will they be spending more time outdoors, and have the windows open to allow in outdoor air.

So I thought I’d do my own little exercise in How to Lie With Statistics (Amazon).  Because it takes about 10 days for an infection to filter into the data, I plotted new COVID-19 cases against the temperature at Reagan National, lagged 10 days.  Changes in that are my proxy for changes in temperature statewide.  And if course, in the spirit of the book, I started the graph at the nearest round numbered day that made the data look good.  (But not suspiciously good.) And came up with this (same as at the top of the posting):

The shame of it is that if there actually is any correlation between weather and disease spread, it’s going to get lost in all the other noise.  Noise, meaning all the other factors that can cause an outbreak in a population with little native immunity to the disease.  Things like (e.g.) opening up the bars, sending kids back to school, and whatnot.

But for this one brief shining moment, hey, that graph looks pretty good.  Until such time as it doesn’t.  I dedicate it to Darrell Huff.

A deeper dive into temperature, humidity, and outdoor restaurant seating.

What brings all this up is a question that a friend asked this morning.  It had to do with adding (electric) heating devices to outdoor restaurant seating.  Presumably, so that restaurants can continue to use that seating on into fall and winter.  (Without risking burning down an overhead canopy with a natural gas or propane heater.)

But, by raising temperatures, those outdoor heaters “dry out” the air (reduce the relative humidity).  Doesn’t that raise the risk of COVID-19 transmission?

Answering that turned out to be a little bit complicated.  But the bottom line is clear:  No.  For two distinct reasons, No.  Outdoor heaters, to preserve outdoor dining into the cold season, should almost certainly be a winner, from the standpoint of preventing spread of COVID-19.

Let’s get to it.

What does the research say regarding humidity and COVID-19 spread?

And, in the true spirit of How to Lie With Statistics, who says so, and how do they know?

And the answers are:  All other things equal, higher relative humidity reduces the spread of COVID-19.  Virtually everyone who has studied this seriously says so.  And  mostly, they know that from epidemiological studies — statistical analysis of enough countries and areas so that the “random” factors wash out, and you can observe any systematic differences related to the weather.

I’m just going to splatter a few references on the page here.  This was enough to give me the gist of it.

Let me cite the key passage from that last one:

"A 1 °C increase in temperature was associated with a 3.08% ... reduction in daily new cases ... a 1% increase in relative humidity was associated with a 0.85% ... reduction in daily new cases ...".

Let me just take those numbers and run with them, without worrying about just how accurate or inaccurate they are.  The clear implication is that heating air, and reducing its relative humidity, have offsetting effects in likelihood of COVID-19 transmission.

Adding patio heaters to outdoor dining makes the outdoor dining area a tiny bit safer, from the standpoint of spreading COVID-19.

I’m going to take as my “typical Virginia winter day” a case where the outdoor air temperature is 45F, at around 50% relative humidity.  If I raise that to 65F (a comfortable temperature), based on this calculator (or ( ), that will reduce the relative humidity by 30 percentage points.

So, for my “typical Virginia winter day”, adding patio heaters to outdoor restaurant seating will increase temperatures 20F (11C), and reduce relative humidity by 30%.  If I use the parameter estimates in the box in the prior section, the net impact of that, on ease of transmitting COVID-19, is calculated as:

(11F x -3.08%) + (30 percentage points x 0.85%) = ~ -8%.

In other words, based on that one set of estimates, heating the air and reducing the relative humidity, on my “typical Virginia winter day”, actually makes the outdoor dining area tiny bit safer.

Customer “behavioral response” is the big unknown

Notice that I keep saying “outdoor dining area”, and not outdoor dining.  That’s because that tiny little minus-eight-percent change calculated above could easily be swamped by changes in the behavior of the diners. 

For the moment, let’s take the Japanese research (in the Atlantic cite above) as true.  That means that public activities in indoor spaces are twenty times riskier than those same activities in outdoor spaces.  As above, let me just run with that, without questioning it.  (Even though, if you dig into it, it’s based on just over 100 cases of disease transmission.)

First point:  If the number of restaurant-goers does not change, anything that induces a person to dine outdoors instead of indoors is unambiguously good.  By inducing them to sit outdoors, you reduce their risk 20-fold.  Conversely, for every diner who goes inside because the outdoor dining is too cold, that increases their risk 20-fold.

The clear upshot is that to the extent that patio heaters induce diners to eat outdoors, that is a major public health benefit.

Second point:  Inducing people to dine outdoors, rather than stay at home, probably raises their COVID-19 risks a bit.  And so, if heating the outdoor dining area draws huge crowds that would otherwise have stayed home, the sheer increase in (low-risk outdoor diners) might tip the balance.

(If there are any restaurant owners reading this, I can guess what you think about that “huge crowds” phrase.  You should be so lucky as to be troubled by huge crowds of outdoor diners.)

I don’t know about you, but just walking around the past couple of days, I sure haven’t seen anything that I’d classify as huge crowds of outdoor diners.  And the weather has been just about as good as it gets around here.  So I’m just dismissing this possibility.


Near as I can tell, heating up an outdoor dining area (and so reducing the relative humidity of the air there) is more-or-less a wash, from the COVID-19 safety of that outdoor dining area.  It very nearly does not matter.

The real public health benefit of heating those outdoor dining areas is that it will keep people from dining indoors.  Indoor dining creates far higher risks of COVID-19 transmission than outdoor dining does.  (Which is the whole reason why state reopenings are written as they are.)  And that effect — plausibly a 20-fold difference in transmission risk between indoor and outdoor settings — is what makes heating outdoor dining areas a significant plus from a public health standpoint.