Friday, August 02, 2019

How Large Is the Urban Heat Island Effect?

Does the urban heat island effects influence global temperature trends?

Roy Spencer claims, without proof, that they do. Berkeley Earth (BEST) found they do not, after having actually examined the global thermometer set six ways from Sunday with a skeptical eye:
"The Urban Heat Island effect is real. Berkeley’s analysis focused on the question of whether this effect biases the global land average. Our UHI paper analyzing this indicates that the urban heat island effect on our global estimate of land temperatures is indistinguishable from zero."

"Influence of Urban Heating on the Global Temperature Land Average using Rural Sites Identified from MODIS Classifications," Wickham et al., Geoinfor Geostat: An Overview 2013, 1:2
And indeed, that's been the determination for some decades; here's a 1990 paper from Nature:
"The results show that the urbanization influence in two of the most widely used hemispheric data sets is, at most, an order of magnitude less than the warming seen on a century timescale."

"Assessment of urbanization effects in time series of surface air temperature over land," P. D. Jones et al, Nature, volume 347, pages 169–172 (1990).
Is there a way to estimate the influence of the UHI? Here's my attempt.

At the Astrobiology conference I was at a couple of weeks ago, a speaker said that on Earth cities occupy 0.2% of the globe's surface area. (It was in the context of the possibility we might see lights on an exoplanet's surface.) I also learned a new word then: "ecumenopolis" -- a planet whose entire surface is covered by one gigantic city.

If there are no UHIs, i.e. no net UHI, then the global mean average temperature (GMAT) is

Now let's say there is a UHI, the same in all cities, and those cities occupy a certain area. Then the GMAT would be something like this weighted average:

(Apologies for the different font sizes.) So the difference in GMAT between a world with a UHI and one with no UHI is

The ratio on the right-hand side is 0.002, and the urban heat island effect is, what, about 5 K (=9°F)? Then

That's your UHI. The trend will change even slower than this, per decade.

Even if the area of all the cities doubles, it will still be only about 0.02 K << global warming of 1 K.

Just a back of the envelope calculation, but it suggests the UHI is indeed small-to-negligible. Because the Earth is vaaaaaaaaaaaaaast.


Layzej said...

The trend in RSS (Satellite) is greater than the trend in BEST, but not by much. You'd think if the surface station record was contaminated by UHI you would be able to see it.

David Appell said...

Nice point.

Layzej said...

"ecumenopolis" sounds nightmarish by the way.

William M. Connolley said...

But thermo dist is uneven. There are prob more in / near cities than your simple calc suggests.

David Appell said...

Of course it's uneven. (Is it uneven when averaged over the globe?) This is just a back of the envelope calculation. Even if the UHI is 10 C, or the area of cities is twice as large, that's only a GMAT change of 0.02 C.

If the area of cities grows exponentially and doubles in, say, 25 years (seems too quick, but let's go with this), and the UHI is 10 C, that's only an addition to the trend of ln(2)*(10 C)/25 years = 0.006 C/decade. Compared to the current 0.2 C/dec.

Victor Venema said...

William is right. The problem is that "cities occupy 0.2% of the globe's surface area", while the number of stations in cities is much larger.

If the percentage of the Earth that is urban is equal to the percentage of urban stations there would be no problem. Then we would sample the warming of the planet right.

It would also easily be possible for urban stations to record too little warming: If the urban heat island effect when and where the measurement started was larger than the urban heat island at the current location. Many stations started in the centre of cities, making measurements in a window screen on a North wall, while now we take much more care that stations are well sited because we now know better know that it matters and we require more accurate observations.

Victor Venema said...

The trend at the US climate reference stations (USCRN), which are sited well away from cities, is larger than the trend in the homogenised US climate average computed from all climate stations. Another sign that if the US climate debate were about the truth, they would not be talking about urbanisation that much.

David Appell said...

Victor wrote:
"The problem is that "cities occupy 0.2% of the globe's surface area", while the number of stations in cities is much larger."

But the GMST is not just the average of all stations, it's the average of all cells, where each cell has the same area and a cell's temperature average is the average of all stations in that cell. No?

Victor Venema said...

The same would go inside a box. The urban stations would be taken to be representative for the rural region around the city.

But yes, this is one reason you cannot simply make the statistics on a per station basis, but have to compute the area average temperature formally. The boxes which have most urban influence are likely also the boxes with many stations. The weight of urban stations in computing a global mean temperature is thus lower than average. The big weights go to isolated stations in the middle of nowhere.