Meta note: don’t see many blog posts as detailed as OP with seemingly mundane details. It assumes almost nothing of the user. So many posts would just say “install this QGIS plugin” with no mention of the pip calls, etc.
Not sure if this comes from search engine optimisation or years of experience consulting to clients that know nothing :)
Anyway, good work Mark. Another of your posts I’ll bookmark.
> The United Nations (UN) believes there are 4B buildings on Earth. This week, a dataset called "GlobalBuildingAtlas" (GBA) was published by researchers at the Technical University of Munich (TUM) that attempts to estimate this number at being closer to 2.75B.
If we painted the roofs on all of them white, by how much would the temperature of the planet drop?
There was a good study on this a few years ago that ran the numbers on this and landed on white paint for residential homes as the best option, for a few reasons, if I remember correctly:
- Installation, maintenance and transmission costs are lower when solar is aggregated on farms
- Solar offsets air conditioning, but that moves the heat outside. White roofs reduce the need for AC, which helps significantly with urban heat scenarios
I'm now wondering if painting a roof white is better than covering the same roof with solar panels and using that to drive air conditioning (in the house). My intuition is that painting it white must be vastly more environmentally friendly, although it probably doesn't reduce the temperature very much compared to aircon.
Back-of-the-envelope
- Road area. World road length ≈ 60–70 million km. Using an average paved width of ~8–10 m ⇒ area ≈ (0.9–1.3)×10¹² m². Earth’s surface is 5.1×10¹⁴ m², so roads cover ~0.09–0.13% of the planet.
- Albedo change. Dark asphalt is ~0.05–0.10. “White” coatings can push toward ~0.4–0.6 (fresh), but weathering quickly dulls them. So a plausible Δalbedo for roads is +0.2 to +0.5.
- Global albedo change. Δα_global ≈ (road fraction) × (Δalbedo_road) ≈ (0.001)×(0.2–0.5) ≈ +0.0002 to +0.0005.
- Radiative forcing. Globally averaged incoming sunlight ≈ S₀/4 ≈ 340 W m⁻². Forcing from an albedo change is ΔF ≈ −Δα_global × 340 ≈ −0.07 to −0.17 W m⁻².
- Temperature response. Using a standard sensitivity ~0.8 °C per W m⁻² (≈3 °C per CO₂ doubling): ΔT ≈ −0.05 to −0.14 °C at equilibrium.
White roads could potentially be blinding, but yeah something lighter than what we do currently could be very worthwhile. It'd have much higher nighttime visibility too.
Painting roofs (or even all man-made surfaces) white wouldn't do much globally at all, but it might reduce the urban heat island effect. But probably (speculating a bit here) not quite so much as creating more green spaces (or green roofs, for that matter).
I would guess not that much - certainly fractions of a degree - because the proportion of the planet's surface is still relatively low, and as much as 60% of that reflected light would be scattered and absorbed by the atmosphere.
It'd have a much bigger impact if all those roofs had solar panels, and the resulting electricity was used to replace carbon-emitting energy sources.
AI is responsible for 60% of all internet traffic this morning, OP is the other 40%.
But on a serious note this is an interesting project highlighting the sheer volume of power usage across the world. .. I'm also curious given the data and power usage, if AI assisted heatmap is actually more accurate than one developed by downloading a single dataset of high resolution satellite imagery at night, performing algorithmic analysis to draw heatmap based on light output or !maybe even simple image processing downsampling the image and shifting white-blue to red?
70% of daytime RGB sat imagery is covered by clouds. I'm not sure how easy it would be to spot if clouds were covering a city's lights at night.
I've only seen Maxar publish one night time image and that was of Dubai. I suspect smaller buildings in not so well lit areas could end up getting missed out.
SAR imagery would work well for seeing at night and through clouds but I'm not sure what the state of AI building footprint detection is with SAR atm.
The 151M Buildings for the US is a bit of a surprise for me.
I found it interesting the disparity between the number of buildings in India and the rest of the world (3 times the next country). And the close parity in number between China and the US.
From what I've seen in news reports, China has built a lot of tower blocks that are 10s of floors, rather than the 4-5-floor buildings I saw when I worked in and travelled around India.
India has ~4x the population of the US so the ratio of buildings isn't much of a surprise.
Is it possible to calculate from that data the total surface area of all roofs? I think there is paint that converts solar energy into radiation that goes straight back into space. I wonder if using that paint on a significant percentage of buildings would matter.
Are you perhaps thinking of radiative cooling, e.g. paint that reflects close to 100% of all light, excluding the spectrum at which room-temperature objects radiate heat, resulting in a net cooling effect?
They reflect almost all sunlight, so the surface doesn’t heat up.
At the same time, they emit thermal radiation in a specific infrared range (around 8–13 micrometers) that can pass through the Earth’s atmosphere and escape into space.
Such paint can make surfaces cooler than the surrounding air, even in sunlight.
So it doesn’t convert solar energy into another form — it selectively reflects and emits infrared radiation that sends heat directly into space.
I see it says Corsair so cannot tell what exact model it is but I did relatively similar thing.
Reason being if you keep the load under certain wattage the PSU will run in passive cooling mode. My rig will never reach 50% of what Corsair SF750 Platinum can deliver and not mention in normal light load circumstances. It spins up its fans only when the load reaches ~300W or so.
Some people are very anal about any kind of noise coming out of their rigs. I personally undervolt everything to keep the fans at bay/minimum and having extra legroom in PSU department helps a lot
Same. Part of the background info on this is PSUs are usually most efficient at (or just under) 50% load. They are also more efficient at 240 V than 120 V, if you have the circuit. So the real efficiency can end up varying significantly, depending how you use it. The efficiency is not usually much to write home about in terms of your electric bill, but it does help drive the ability for the PSU to cool itself passively.
I remember reading something interesting, in a good way, but I'm not sure whether it's fully accurate: That the majority of the planet (land, not counting water) is still unpopulated by people.
Depends on how you measure it. Do you subdivide all land in 1 cm squares and only count the ones with somebody standing on them as populated? Then almost all land is unpopulated. With 8.2 billion people in the world, each one occupying a 0.5x0.5 meter square, you have 2.05e9 m^2 populated land, out of Earth's total land area ~1.5e14 m^2 [1]. That's less than 0.0014%.
As you increase the size of your subdivisions, the unpopulated fraction goes down. In the limit of all land being in just one subdivision, it's obviously 0.
These type of calculations don't really work for things like skyscrapers, high-rise apartment complexes, or the slum-towns of Asia and South America, and conversely for very sparsely populated regions like the fringes of the Sahara. Antarctica has people too but it doesn't make any sense to count it as "populated".
I think we can just draw an outline around cities, towns, and villages and count that area as populated, no matter how many people are actually in there.
For long stretches of land where there's just a single highway going through and a few gas stations, motels and shacks, totaling maybe <100 people for many miles around, we can count that land as unpopulated.
Isolated islands with indigenous/uncontacted tribes, not sure.. We may have to consider their population growth and see if they're in "equilibrium" with "nature" and so on.
A simple glance at Google Earth shows large parts of Russia as clearly unpopulated/undelevoped, then there's the deserts of China and Africa, and some natural land in the Americas, and finally of course there's Antarctica, until the Elder Things awaken.
The US has only used 25% of its land. Overture published a land use dataset a while back that could go some way to verify how much land on earth is urban, covered in forest, etc.. That might only need a single SQL statement given how they structured their data.
I remember the same thing. I don't have any sources, but I've come to understand that a lot of land is on hills/mountains and is therefore an inconvenient/dangerous place to build.
I think the "overcrowded planet" scenario is not very plausible, except if we're talking about resources. There is plenty of land for everyone, especially if we start to construct efficiently.
> There is plenty of land for everyone, especially if we start to construct efficiently.
I was afraid someone will get it the wrong way: I hope it STAYS that way; that the majority of the planet remains unpopulated, but humans learn to make better use of the space we're already occupying.
Does Agriculture land count as "populated" or not? If not, most of the land is empty in most countries (even in Japan). A quick search shows Singapore "built-up" area around a third of its total surface.
The only exception might be the vatican, if you consider it a country.
Kind of crazy to think that there are almost half as many buildings as people and yet me and my wife can't even afford the deposit to buy a single one on a mortgage after 10 years of saving, hustling hard, taught myself to code, got a university degree, worked in the tech sector, near the forefront of all the hot tech tends... Meanwhile, all this time, I'm told I'm privileged. I must be the victim of some kind of massive PsyOp conspiracy.
Most of those buildings are so poor you probably wouldn’t even consider one outside the top 10% of the world for free assuming you want a first world kind of accommodation. Source: I’ve seen the world.
In my country (Australia), the cheapest I can get is like $250k USD equivalent... But that's a shack in the middle of the desert in some tiny ghost town. Even in a normal small town, houses are like $350k USD minimum. In capital cities, it's like $500k USD minimum for a tiny apartment... Top salaries for top senior software devs are like $130k USD before tax, taxed at about 40%... So take home pay is like $78k USD net income. Rent is like $22k per year minimum. Impossible to get fixed-rate mortgages; banks only only do variable rate... 20% minimum deposit. This is in a country with a LOT of spare land which is scarcely populated.
IMO the real problem is that the monetary system creates asymmetries which create monopolies, which force everyone to participate in 'the economy' through rigid, inefficient, bureaucratic structures. People are literally not allowed to service each other, even through they have the time and capacity to so; simply because they do not have access to a specific currency which they are forced to use...
That same currency is not universally scarce though; it's basically handed out to their biggest corporate competitors in large quantities via government contracts and from banks in the form of low interest loans. It creates a kind of fictitious planned economy which becomes increasingly inefficient and bureaucratic by the day; the money is the only quantity which is increasing (and only for some people), almost every other economic metric is decreasing. The real economy is hollowing out.
Meta note: don’t see many blog posts as detailed as OP with seemingly mundane details. It assumes almost nothing of the user. So many posts would just say “install this QGIS plugin” with no mention of the pip calls, etc.
Not sure if this comes from search engine optimisation or years of experience consulting to clients that know nothing :)
Anyway, good work Mark. Another of your posts I’ll bookmark.
> The United Nations (UN) believes there are 4B buildings on Earth. This week, a dataset called "GlobalBuildingAtlas" (GBA) was published by researchers at the Technical University of Munich (TUM) that attempts to estimate this number at being closer to 2.75B.
If we painted the roofs on all of them white, by how much would the temperature of the planet drop?
Probably not as much as if we “painted” them black with solar panels and used the resulting electricity to displace fossil fuel burning.
There was a good study on this a few years ago that ran the numbers on this and landed on white paint for residential homes as the best option, for a few reasons, if I remember correctly:
- Installation, maintenance and transmission costs are lower when solar is aggregated on farms - Solar offsets air conditioning, but that moves the heat outside. White roofs reduce the need for AC, which helps significantly with urban heat scenarios
A quick search yields a UCL study, which supports the lower claim: https://phys.org/news/2024-07-roofs-white-city.html
A good percentage of Pakistan did this recently and removed 35% of demand off of their grid https://www.ted.com/talks/jenny_chase_solar_energy_is_even_c...
5 replies →
I'm now wondering if painting a roof white is better than covering the same roof with solar panels and using that to drive air conditioning (in the house). My intuition is that painting it white must be vastly more environmentally friendly, although it probably doesn't reduce the temperature very much compared to aircon.
9 replies →
there is a cool project that explores this topic: https://coolcolors.lbl.gov/
I think we should do more albedo engineering. like white roads and car roofs.
Back-of-the-envelope - Road area. World road length ≈ 60–70 million km. Using an average paved width of ~8–10 m ⇒ area ≈ (0.9–1.3)×10¹² m². Earth’s surface is 5.1×10¹⁴ m², so roads cover ~0.09–0.13% of the planet.
- Albedo change. Dark asphalt is ~0.05–0.10. “White” coatings can push toward ~0.4–0.6 (fresh), but weathering quickly dulls them. So a plausible Δalbedo for roads is +0.2 to +0.5.
- Global albedo change. Δα_global ≈ (road fraction) × (Δalbedo_road) ≈ (0.001)×(0.2–0.5) ≈ +0.0002 to +0.0005.
- Radiative forcing. Globally averaged incoming sunlight ≈ S₀/4 ≈ 340 W m⁻². Forcing from an albedo change is ΔF ≈ −Δα_global × 340 ≈ −0.07 to −0.17 W m⁻².
- Temperature response. Using a standard sensitivity ~0.8 °C per W m⁻² (≈3 °C per CO₂ doubling): ΔT ≈ −0.05 to −0.14 °C at equilibrium.
White roads could potentially be blinding, but yeah something lighter than what we do currently could be very worthwhile. It'd have much higher nighttime visibility too.
4 replies →
Teslas will start crashing like there’s no tomorrow.
1 reply →
Painting roofs (or even all man-made surfaces) white wouldn't do much globally at all, but it might reduce the urban heat island effect. But probably (speculating a bit here) not quite so much as creating more green spaces (or green roofs, for that matter).
I would guess not that much - certainly fractions of a degree - because the proportion of the planet's surface is still relatively low, and as much as 60% of that reflected light would be scattered and absorbed by the atmosphere.
It'd have a much bigger impact if all those roofs had solar panels, and the resulting electricity was used to replace carbon-emitting energy sources.
Paris and other climate accords have much ado about fractions of a degree, leading me to believe that that's highly consequential.
AI is responsible for 60% of all internet traffic this morning, OP is the other 40%.
But on a serious note this is an interesting project highlighting the sheer volume of power usage across the world. .. I'm also curious given the data and power usage, if AI assisted heatmap is actually more accurate than one developed by downloading a single dataset of high resolution satellite imagery at night, performing algorithmic analysis to draw heatmap based on light output or !maybe even simple image processing downsampling the image and shifting white-blue to red?
70% of daytime RGB sat imagery is covered by clouds. I'm not sure how easy it would be to spot if clouds were covering a city's lights at night.
I've only seen Maxar publish one night time image and that was of Dubai. I suspect smaller buildings in not so well lit areas could end up getting missed out.
SAR imagery would work well for seeing at night and through clouds but I'm not sure what the state of AI building footprint detection is with SAR atm.
Fascinating. So no such satellite data exists. Thanks for the insight, makes the linked write up much more satisfying.
The 151M Buildings for the US is a bit of a surprise for me.
I found it interesting the disparity between the number of buildings in India and the rest of the world (3 times the next country). And the close parity in number between China and the US.
From what I've seen in news reports, China has built a lot of tower blocks that are 10s of floors, rather than the 4-5-floor buildings I saw when I worked in and travelled around India.
India has ~4x the population of the US so the ratio of buildings isn't much of a surprise.
Is it possible to calculate from that data the total surface area of all roofs? I think there is paint that converts solar energy into radiation that goes straight back into space. I wonder if using that paint on a significant percentage of buildings would matter.
Are you perhaps thinking of radiative cooling, e.g. paint that reflects close to 100% of all light, excluding the spectrum at which room-temperature objects radiate heat, resulting in a net cooling effect?
E.g. [Revolutionary Paint: How to Make Surfaces Stay Cool in the Sun](https://youtu.be/dNs_kNilSjk)
Thanks for the video!
Yes, from chatgpt:
These are so-called “radiative cooling paints.”
Here’s how they work:
They reflect almost all sunlight, so the surface doesn’t heat up.
At the same time, they emit thermal radiation in a specific infrared range (around 8–13 micrometers) that can pass through the Earth’s atmosphere and escape into space.
Such paint can make surfaces cooler than the surrounding air, even in sunlight.
So it doesn’t convert solar energy into another form — it selectively reflects and emits infrared radiation that sends heat directly into space.
Ok a bit off topic but isn't a 1200W PSU overkill for this system? For a 9950X and 1080, 500W seems plenty.
I'm hoping to use most of this system for the next 10 years. At some point, I want to add some beefy GPUs to it when I get back into 3D again.
I see it says Corsair so cannot tell what exact model it is but I did relatively similar thing. Reason being if you keep the load under certain wattage the PSU will run in passive cooling mode. My rig will never reach 50% of what Corsair SF750 Platinum can deliver and not mention in normal light load circumstances. It spins up its fans only when the load reaches ~300W or so.
Some people are very anal about any kind of noise coming out of their rigs. I personally undervolt everything to keep the fans at bay/minimum and having extra legroom in PSU department helps a lot
Same. Part of the background info on this is PSUs are usually most efficient at (or just under) 50% load. They are also more efficient at 240 V than 120 V, if you have the circuit. So the real efficiency can end up varying significantly, depending how you use it. The efficiency is not usually much to write home about in terms of your electric bill, but it does help drive the ability for the PSU to cool itself passively.
I remember reading something interesting, in a good way, but I'm not sure whether it's fully accurate: That the majority of the planet (land, not counting water) is still unpopulated by people.
Can someone verify?
Depends on how you measure it. Do you subdivide all land in 1 cm squares and only count the ones with somebody standing on them as populated? Then almost all land is unpopulated. With 8.2 billion people in the world, each one occupying a 0.5x0.5 meter square, you have 2.05e9 m^2 populated land, out of Earth's total land area ~1.5e14 m^2 [1]. That's less than 0.0014%.
As you increase the size of your subdivisions, the unpopulated fraction goes down. In the limit of all land being in just one subdivision, it's obviously 0.
[1] https://en.wikipedia.org/wiki/Population_density
These type of calculations don't really work for things like skyscrapers, high-rise apartment complexes, or the slum-towns of Asia and South America, and conversely for very sparsely populated regions like the fringes of the Sahara. Antarctica has people too but it doesn't make any sense to count it as "populated".
I think we can just draw an outline around cities, towns, and villages and count that area as populated, no matter how many people are actually in there.
For long stretches of land where there's just a single highway going through and a few gas stations, motels and shacks, totaling maybe <100 people for many miles around, we can count that land as unpopulated.
Isolated islands with indigenous/uncontacted tribes, not sure.. We may have to consider their population growth and see if they're in "equilibrium" with "nature" and so on.
A simple glance at Google Earth shows large parts of Russia as clearly unpopulated/undelevoped, then there's the deserts of China and Africa, and some natural land in the Americas, and finally of course there's Antarctica, until the Elder Things awaken.
The US has only used 25% of its land. Overture published a land use dataset a while back that could go some way to verify how much land on earth is urban, covered in forest, etc.. That might only need a single SQL statement given how they structured their data.
I have some analysis around that topic in the middle of this post: https://tech.marksblogg.com/overture-land-cover.html
I remember the same thing. I don't have any sources, but I've come to understand that a lot of land is on hills/mountains and is therefore an inconvenient/dangerous place to build.
I think the "overcrowded planet" scenario is not very plausible, except if we're talking about resources. There is plenty of land for everyone, especially if we start to construct efficiently.
> There is plenty of land for everyone, especially if we start to construct efficiently.
I was afraid someone will get it the wrong way: I hope it STAYS that way; that the majority of the planet remains unpopulated, but humans learn to make better use of the space we're already occupying.
Does Agriculture land count as "populated" or not? If not, most of the land is empty in most countries (even in Japan). A quick search shows Singapore "built-up" area around a third of its total surface.
The only exception might be the vatican, if you consider it a country.
Kind of crazy to think that there are almost half as many buildings as people and yet me and my wife can't even afford the deposit to buy a single one on a mortgage after 10 years of saving, hustling hard, taught myself to code, got a university degree, worked in the tech sector, near the forefront of all the hot tech tends... Meanwhile, all this time, I'm told I'm privileged. I must be the victim of some kind of massive PsyOp conspiracy.
Most of those buildings are so poor you probably wouldn’t even consider one outside the top 10% of the world for free assuming you want a first world kind of accommodation. Source: I’ve seen the world.
You can buy a small house in rural America for $25,000
In my country (Australia), the cheapest I can get is like $250k USD equivalent... But that's a shack in the middle of the desert in some tiny ghost town. Even in a normal small town, houses are like $350k USD minimum. In capital cities, it's like $500k USD minimum for a tiny apartment... Top salaries for top senior software devs are like $130k USD before tax, taxed at about 40%... So take home pay is like $78k USD net income. Rent is like $22k per year minimum. Impossible to get fixed-rate mortgages; banks only only do variable rate... 20% minimum deposit. This is in a country with a LOT of spare land which is scarcely populated.
2 replies →
Rural Russia as well. But you will probably have to get your water from a well and poop into a latrine.
1 reply →
Many buildings will be commercial, industrial, military etc. That said one building can house 1000 people.
Oh, poor poor jongjong... my sympathies!
Chatting with ChatGPT:
Humans are in oversupply currently - for what needs doing.
And as we all know, products in oversupply lose their value.
The algorithms that drive the world are optimised for money and capital, humans are just one product in that structure.
Enjoy the ride.
IMO the real problem is that the monetary system creates asymmetries which create monopolies, which force everyone to participate in 'the economy' through rigid, inefficient, bureaucratic structures. People are literally not allowed to service each other, even through they have the time and capacity to so; simply because they do not have access to a specific currency which they are forced to use... That same currency is not universally scarce though; it's basically handed out to their biggest corporate competitors in large quantities via government contracts and from banks in the form of low interest loans. It creates a kind of fictitious planned economy which becomes increasingly inefficient and bureaucratic by the day; the money is the only quantity which is increasing (and only for some people), almost every other economic metric is decreasing. The real economy is hollowing out.
2 replies →