The world’s surface temp is ~14°C, but it’s very likely that the human population’s (outdoor) temp is higher, because lots of people live near the equator and few live near the poles. The same can be said for individual countries: there are lists that tell us a nation’s yearly temp, but I’m fairly certain that they calculated it by averaging every square kilometer and not accounting for pop density in each sqkm. That explains why millions can live in, say, Russia, in a -5.35°C climate.
Is there any study on people’s actual outdoor average temp on a scale of countries? If yes, what would be the keyword for that mouthful? If not, can we build a simple model/algorithm to calculate it based on existing maps’ pop density data?
That is a very interesting question. A quick Google led me to this:
Here’s a similar list of populations:
One could easily enough copy those data tables into a spreadsheet matching on country name and produce a weighted average that way.
It would not be perfect, as e.g. the US population or the Chinese population may be skewed towards living hotter or colder than their respective whole-country average. But you’ll be a lot closer than the 14C number your starting with.
I’ll also point out that the top 20-ish countries by population cover more than half the world population. So you really only need to do the first 20 rows of the population table to get a pretty decent approximation for the whole world. By 30 rows you’re getting very close; no need to do all 200.
Looking at the map in the temperature article, most of the red really hot ones are low population except Indonesia. Lots of population in the dark orange ones, but also lots of population in the light blue gray ones. The light orange ones of intermediate temp are relatively few and relatively low-population.
So …
I’ll suggest the dark orange and the light blue gray pretty well are a toss-up in terms of headcount and represent a very large slice of humanity. So in terms of temp they offset each other. Adding in the other colors leans the headcount and hence the average a bit towards warmer than colder.
Which gives me a Feynman estimate / visual estimate that a complete proper calc would come out real close to 20C.
Perhaps better to consider the temperatures of urban areas - where much of the population live, fairly consistent with surrounding inhabited countryside, and localized enough to eliminate including the more extreme polar or desert regions. (Death Valley and the Sierras are not - usually - representative of California, I assume)
And Wikipedia does thankfully have a list of average temperatures of 470 cities. Would be pretty easy to scrape the data and get population for each:
For the sake of the exercise, matching that list to lists of largest cities and populations (ie incomplete) gives the following results.
Africa 23.46C (49 cities)
Asia 22.39C (25 cities)
If the mood takes me this evening might extend that list to the other continents and perhaps more cities per continent.
That’s right. I believe the Chinese live in a much warmer climate than displayed in the wiki link, because a huge chunk of cold Tibetan highland is sparsely populated. Same goes with Alaska.
I read that cities contribute ~56% of pop. So basically a 50/50 split. We should note that because of the heat island effect, non-urban areas are ~2°C lower (surface only, not counting pop). I’d argue that since cities usually congregate on the warmer plots in the 1st place, while the remaining 44% rural pop are spread over much colder lands, the average should be skewed a bit lower to reflect the situation. In other words, if we calculate using the city pop route and get X, the final figure might be X-1.25, or X-1.5.
A nice catch! I assume you intend to do it automatically? While I’m a noob on IT, I’d love to learn some 101-lesson. Like, are there software whose purpose is to help you scrape? In this particular case, what commands would you input?
Great, have a good day then! We’re halfway there already, IMO.
Now, my idea when I posted the OP was something like this: divide all the world’s land surface into 10x10km plots, so 1.489.400 pieces in total. Divide global pop into dot, 1000 people each, so 8.031.316 dots total. Put the dots onto the pieces - something like a combination of this and this link. Retrieve the climate data for each piece, and do the last step of simple weighted calc. I believe with all the satellites, census, and whatnot nowadays, we have more than enough data to do all of this. The problem is that it’s a bit too scientific and labor-intensive.
BUT! This seems to be a pretty important aspect, and shouldn’t it be covered by nations’ research already? The “POT” (Population’s Outdoor Temp) surely is related to / can help with many sociology and environmental issues.
Based on the listing from @AlsoNamedBort
Africa 23.41C (110 cities)
Asia 21.68C (97 cities)
North America 16.13C (98 cities)
weighted average: 21.57C (305 cities)
combined population: 857mil
Note those 305 cities represent just over 10% of the human population. 857M out of 8031M humans per World Population Clock: 8 Billion People (LIVE, 2023) - Worldometer (worldometers.info).
As to whether people build cities in warmer or cooler than average places within their country…
I’ll point out that until the 1950s the vast bulk of US population lived where it snowed every year. No, they’re not on mountaintops, but they also didn’t live in the hot parts if that could be avoided. Times are changing of course and at least in the USA the last 50 years have been a net migration towards warmer. But there’s still a heck of a lot of Americans anc Chinese that live in cold cities.
Yes. As the saying about statistics goes - “The average person has one breast and one testicle.”
Getting an “average” for USA temperature/population is less meaningfull than, say, Germany or even Canada. China too runs from temperate to tropical.
My other point was that while cities may be less than 100% of the population, a significant number of the rest live in proximity to those urban centers; so you are probably catching more than 80% of the population within, say, 200 km of major urban centers and thus sharing the climate generally. (For various definitions of “major urban center”)
For example, 100km of London, Manchester, Birmingham, Glasgow/Edinburgh - what proportion of British population would that capture?
The population covered is likely much higher because the city population was used, not the metropolitan or hinterland.
E.G. St. Louis In 2020, the city proper had a population of 301,578, while its bi-state metropolitan area, which extends into Illinois, had an estimated population of over 2.8 million.
In the analysis above St Louis is based on the 301k.
Especially in the USA, the “city” does not truly encompass the urban sprawl. Even in Canada, where suburban villages have been consolidated into the central city, the sprawl has outgrown that concept. Metro Toronto went from 13 towns to 13 boroughs to 7 and then 1 big city with no subdivisions - but the suburbs have badly outgrown even that. The USA seems to lack the impetus to consolidate tiny municipalties leaving urban governments fragmented, but distributing things like transit to separate authorities less dependent on one municipality, it seems.
Still waiting for your excellent analysis of Europe and SA, but with a reasonable assumption that Europe is slightly colder than NA, and SA is slightly warmer, I came to a world cityPOT number of ~21.3°C. Combine that with another speculation about colder rural, and the final POT is ~20°C, just as @LSLGuy predicted. Holy thermometer!
Which raised my interest in the mentioned Feynman estimate technique. However, my search only yielded “Fermi problem”. Could you help a bit with this, LSL?
That’s easy. I’d misremembered the namesake of the technique. It’s a Fermi Estimate, not a Feynman Estimate. D’oh! :forehead_smack:
See here, which I expect you already have:
Although I’ve got a persistent recollection that Feynman was also good at these. Google is not exactly backing me up on that though.
Here’s a pretty approachable explanation of the process: Paint the Earth (xkcd.com)
Based on the listing from @AlsoNamedBort
| Continent | Average Temp C | Cities |
|---|---|---|
| Africa | 23.5C | 110 |
| Asia | 21.7C | 97 |
| Europe | 10.8C | 73 |
| North America | 16.4C | 98 |
| Oceania | 17.4C | 29 |
| South America | 19.9C | 57 |
| Selected Cities | 20.4C | 464 |
It does need to be noted that this listing comprises “selected cities”, not representative or other measure and has a Mercator style perspective as it includes a megapolis like…
| City | Population | Temperature C |
|---|---|---|
| Dikson, Russia | 319 | -11.1 |
| Hamilton, Bermuda | 846 | 22.3 |
| Lake Tekapo, New Zealand | 558 | 8.7 |
… whilst excluding any number of Chinese and Indian cities with populations exceeding 1mil.
Pretty cool, tks. My impression of the technique is that it might actually rely on the number of wildcards. The more unknowns there are, the higher chance that incorrect guesses will ‘cancel’ each other and thus help us arrive at an “accurate” answer. But… in our case, would 40°C still be acceptable? It falls within an order of magnitude…
Wait wait… what? This makes the list at least an order of magnitude less reliable. Thanks for the heads-up, thule! We may have to scrap the city route, though - Indian metropolises, and some Chinese, are hot.
For the sake of the exercise I took the population of all the selected cities in China and India and increased their population ie weighting, by a factor of 10.
So total selected city population in Asia goes from 555mil to 1.9bil.
And the average global temperature of the selected cities falls from 20.4C to 19.7C