I believe this is a truly important piece of work. I hope Dr. Spencer will submit it to a journal. I’m grateful to Dr. Spencer for his email suggesting I post it here. Consider this early peer review. Beat it up, find any errors, and point out flaws, so that he can make it better. – Anthony
by Roy W. Spencer, Ph. D.
UPDATED (12:30 p.m. CST, March 3): Appended new discussion & plots showing importance of how low-population density stations are handled.
ABSTRACT
Global hourly surface temperature observations and 1 km resolution population density data for the year 2000 are used together to quantify the average urban heat island (UHI) effect. While the rate of warming with population increase is the greatest at the lowest population densities, some warming continues with population increases even for densely populated cities. Statistics like those presented here could be used to correct the surface temperature record for spurious warming caused by the UHI effect, providing better estimates of temperature trends.
METHOD
Using NOAA’s International Surface Hourly (ISH) weather data from around the world during 2000, I computed daily, monthly, and then 1-year average temperatures for each weather station. For a station to be used, a daily average temperature computation required the 4 synoptic temperature observations at 00, 06, 12, and 18 UTC; a monthly average required at least 20 good days per month; and a yearly average required all 12 months.
For each of those weather station locations I also stored the average population density from the 1 km gridded global population density data archived at the Socioeconomic Data and Applications Center (SEDAC).
All station pairs within 150 km of each other had their 1-year average difference in temperature related to their difference in population. Averaging of these station pairs’ results was done in 10 population bins each for Station1 and Station2, with bin boundaries at 0, 20, 50, 100, 200, 400, 800, 1600, 3200, 6400, and 50000 persons per sq. km.
Because some stations are located next to large water bodies, I used an old USAF 1/6 deg lat/lon percent water coverage dataset to ensure that there was no more than a 20% difference in the percent water coverage between the two stations in each match-up. (I believe this water coverage dataset is no longer publicly available).
Elevation effects were estimated by regressing station pair temperature differences against station elevation differences, which yielded a cooling rate of 5.4 deg. C per km increase in station elevation. Then, all station temperatures were adjusted to sea level (0 km elevation) with this relationship.
After all screening, a total of 10,307 unique station pairs were accepted for analysis from 2000.
RESULTS & DISCUSSION
The following graph shows the average rate of warming with population density increase (vertical axis), as a function of the average populations of the station pairs. Each data point represents a population bin average for the intersection of a higher population station with its lower-population station mate.
Using the data in the above graph, we can now compute average cumulative warming from a population density of zero, the results of which are shown in the next graph. [Note that this step would be unnecessary if every populated station location had a zero-population station nearby. In that case, it would be much easier to compute the average warming associated with a population density increase.]
This graph shows that the most rapid rate of warming with population increase is at the lowest population densities. The non-linear relationship is not a new discovery, as it has been noted by previous researchers who found an approximate logarithmic dependence of warming on population.
Significantly, this means that monitoring long-term warming at more rural stations could have greater spurious warming than monitoring in the cities. For instance, a population increase from 0 to 20 people per sq. km gives a warming of +0.22 deg C, but for a densely populated location having 1,000 people per sq. km, it takes an additional 1,500 people (to 2,500 people per sq. km) to get the same 0.22 deg. C warming. (Of course, if one can find stations whose environment has not changed at all, that would be the preferred situation.)
Since this analysis used only 1 year of data, other years could be examined to see how robust the above relationship is. Also, since there are gridded population data for 1990, 2000, and 2010 (estimated), one could examine whether there is any indication of the temperature-population relationship changing over time.
This is the type of information which I can envision being used to adjust station temperatures throughout the historical record, even as stations come, go, and move. As mentioned above, the elevation adjustment for individual stations can be done fairly easily, and the population adjustments could then be done without having to inter-calibrate stations.
Such adjustments help to maximize the number of stations used in temperature trend analysis, rather than simply throwing the data out. Note that the philosophy here is not to provide the best adjustments for each station individually, but to do adjustments for spurious effects which, when averaged over all stations, will remove the effect when averaged over all stations. This ensures simplicity and reproducibility of the analysis.
UPDATE:
The above results are quite sensitive to how the stations with very low population densities are handled. I’ve recomputed the above results by adding a single data point representing 724 more station pairs where BOTH stations are within the lowest population density category: 0 to 20 people per sq. km. This increases the signal of warming at low population densities, from the previously mentioned +0.22 deg C warming from zero to 20 people per sq. km, to +0.77 deg. C of warming.
This is over a factor of 3 more warming from 0 to 20 persons per sq. km with the additional data. This is important because most weather observation sites have relatively low population densities: in my dataset, I find that one-half of all stations have population densities below 100 persons per sq. km. The following plot zooms in on the lower left corner of the previous plot so you can better see the warming at the lowest population densities.
Clearly, any UHI adjustments to past thermometer data will depend upon how the UHI effect is quantified at these very low population densities.
Also, since I didn’t mention it earlier, I should clarify that population density is just an accessible index that is presumed to be related to how much the environment around the thermometer site has been modified over time, by replacing vegetation with manmade structures. Population density is not expected to always be a good index of this modification — for instance, population densities at large airports can be expected to be low, but the surrounding runway surfaces and airplane traffic can be expected to cause considerable spurious warming, much more than would be expected for their population density.
“While the rate of warming with population increase is the greatest at the lowest population densities, some warming continues with population increases even for densely populated cities.”
A matter of vertical surface area?
In 1900, our big farm house for few people was deliberately built for maximum winter heat gain, surface area/ sun exposure, as were many others. The north side is flat, south side convoluted, west flat and shaded. No insulation. Adding a garage, a fence, limiting the breezes, made a large part of the yard a solar greenhouse in itself. (One would think solar panels would be bumpy, if this line of thought held water, though I guess they are on the insides.) I can’t calculate the effects of added shade though. I can usually (not this year, no sun) grow February strawberries on the south side while keeping the beer cold in the leftover snow piles on the north side.
High density housing just gets denser per unit. Here people have room to add outbuildings and sprawling semi-rural subdivisions.
The archives here always baffle me, but there was a fascinating old topic with loads of photos, showing the unexpectedly high infra red camera temperatures of buildings, far surpassing the rated” tolerances of roofing materials, for instance. This was not about heat loss, it pertained to the exposed surfaces of buildings, vertical, slanted, flat etc absorbing heat from the sun. Can’t find it.
Re sphaerica (07:32:30) :
sphaerica, you also need to consider that large areas of the world (The BRIC countries for instance) are rapidly growing, as compared to the US and europe, and therefore are showing far more warming, as the UHI effect would be in a stronger part of the curve; look up studies on UHI effect in China for instance, which show about 1 C per century effect to the trend as measured in China. This may be a large part of the reason the US data shows very little warming compared to the global average.
Sir, your method of reporting and the statistical treatment of errors of measurements needs elaboration.
Hugh Roper (05:21:04) :
…the IPCC’s mission is to underline the supposed predominance of CO2 as the leading positive climate forcing, so it is only to be expected that they will play down the role of other potential positive forcings and emphasize negative ones.
Well said, and in case there is any misunderstanding by the other folks here the land use is seen as a negative forcing (~.2 W/m^2) by the IPCC*
The explanation for this, as I understand it, is that replacing forests with crops (deforestation) increases albedo by roughly 10% (7-14% to 15-25%, ref*) – I am not sure on the details here, but I suspect that their storyline might be that UHI has already been adjusted out of the temperature records so there’s no need to include it in their forcings. It could also be that they included it in the land use assessment and the deforestation effect simply outweighs any UHI… so the Net land is still negative.
Putting concrete, structures, etc aside for a moment… the delta for power consumption between 1965 and 2006 is 10.8 TW, with total consumption in 2006 of 15.8 TW. A significant chunk of that energy is heat loss – take the internal combustion engine. It is primarily a heat engine, it creates usable power almost as a side-product (18-20% average efficiency). This means that of the ~5 TW worth of Oil we use, 4 TW is directly lost to heat. Electric power generation (66% loss for coal and nuclear), transmission (7.2%) and usage also exhibit heat loss due to inefficiency on top of direct effects like Air Conditioning.
I think that for a truly accurate assessment of UHI forcing, energy consumption patterns (which, FWIW, I do not believe can be accurately correlated with GDP – ref*) need to be factored in, and I believe that energy patterns could be used to develop a very accurate forcing assessment for this piece of UHI
Note: land use, structures, roads, vegetation etc would still need to be addressed in addition to energy usage patterns… what I’m arguing here is that I think it should be addressed separately.
Now, that being said, what I have been talking about is a fundamentally different idea and approach than Dr. Spencer is doing here. I asked him earlier if there was an discernible difference region to region with his assessment – I think there probably is under the covers – but all-in-all his approach is valid and certainly adequate if not exceeding where climate science is today: which in Economics terms is very macro, not very micro.
That’s my $0.02 at least
@Hugh Roper re: deforestation
Replacing forests with crops most certainly does not increase albedo. I live in a bush lot on the edge of a farm field, I can observe heat shimmering off a ripened wheat field or freshly plowed black earth, I can walk from the shade of the trees into the noticeably warmer green corn field, agriculturization has as great effect on surface temperatures as UHI. Who is studying agriculture heat effect AHE ? Driving from my bush lot to my brothers country home in an open lot in summer I get 2+ degrees C warmer in the open rural lot.
@ur momisugly sphaerica (07:32:21) :
“Dr. Spencer is working purely with population density, not change in population density. … I think a lot of readers are making this mistake, and assuming that the post has to do with what they expect it to address, i.e. changes in population.”
Do you mean you assume that “a lot of readers” have ignored, like you apparently did, the clearly stated fact that Spencer’s figures are based on data from a single year? Please explain:
1) what kind of “change in population density” during that single year you would like Spencer to include in his calculations, and
2) why his results, especially the graph relating pop.density to warming bias, would not apply in ONE location at DIFFERENT times, even though they were calculated by comparing DIFFERENT locations at the SAME time.
As I read the graph, a place whose pop.density was e.g. 1,000 when readings started, gradually increasing to 7,000 today, would have produced temperature readings running hot by 1.5 deg.C at the start, with this bias increasing to 2.2 deg.C today – i.e. a spurious warming trend of 2.2 – 1.5 = 0.7 deg.C . Am I misunderstanding this?
Chris Wright (02:56:25) :
A very nice piece of work. I regard UHI as one of the biggest issues in climate science. Even the Daily Telegraph (whose coverage not so long ago was completely biased in favour of AGW) recently printed a report about UHI. The headline was: “Global warming data skewed by heat from planes and buildings”. How times have changed….
This work confirms something that certainly makes sense: that UHI causes larger temperature changes for smaller growing populations and that the relationship is approximately logarithmic. Of course, the other side of the coin is that, as populations continue to grow, the effect starts to saturate. Once everything is covered with concrete there’s little opportunity for further UHI.
With this in mind, is it possible that the recent lack of warming (since around 1995) could be partly due to UHI reaching saturation in many parts of the world?
Could not agree more! I have been analysing data for Sydney and a smaller city just 100km away, Newcastle. Up until the 1960’s Sydney minus Newcastle was negative. From then the difference grew positive but has plateaued in the last 10 years. Both show a warming that really starts in the 60’s when the skyline of Sydney was undergoing rapid change. Also I looked at rural stations Dubbo and Cessnock both within 500km of Sydney and with data going back to early 20th century. The rural stations show no warming and using the favourite snake oil ingredient of climate science; the linear trend, they show a slight cooling. It is my belief that yes, we have been warming but it is all UHI. Everything is true about warmest decade etc but is UHI not CO2.
I was somewhat surprised to read in the Chicago Tribune that placing weather stations near incinerators, air conditioners and hot jet exhaust actually cools them.
It seems to be more along the lines of hot is cold and cold is hot, a testament to the warmists’ unique ability to hold two contradictory thoughts in their heads at the same time.
http://www.chicagotribune.com/news/ct-met-0228-climate-science-questions-20100302,0,2670932.story
Did the National Oceanic and Atmospheric Administration misplace weather stations and exaggerate warming?
Anthony Watts, a weather forecaster whose Web sites, Watts Up With That and surfacestations.org, have become focal points for climate skeptics, enlisted volunteers across the country to photograph weather stations. Citing NOAA’s own criteria, Watts concluded last year that hundreds of the stations were in poorly located sites, next to parking lots or heating vents or other areas that could inflate temperatures.
“The U.S. temperature record is unreliable,” Watts concluded. “And since the U.S. record is thought to be the best in the world, it follows that the global database is likely similarly compromised and unreliable.”
A new peer-reviewed study by scientists at the National Climatic Data Center, the federal office that tracks climate trends, agrees that problems with the locations of many weather stations are real.
But the temperature records from the poorly located stations cited by Watts actually have a slightly cool bias, not a warm one, according to the review, scheduled to be published in the Journal of Geophysical Research-Atmospheres.
“Fortunately, the sites with good exposure, though small in number, are reasonably well distributed across the country,” the researchers concluded, adding that the “good” or better stations cited by Watts show warming over time similar to NOAA’s overall data.
There also are multiple other surface and satellite measurements of global temperatures, all of which show a warming trend.
Watts says he is writing his own paper, which has yet to be accepted for publication by a scientific journal.
Sorry if I’m derailing here, but for grins I did some back of the envelope calculations regarding forcing from energy use:
Urban Areas accounts for 1.5% of the earth’s land surface, in 2006 total power consumption was 1.58 TW, assuming an average use efficiency of 33% (meaning 66% loss due to heat), and assuming 50% of the usage is in said urban areas (50% of people live in urban areas as of 2008)… we are talking about a forcing in urban areas of 2.37 W/m^2. There would also be a forcing of 5.29 TW spread out among non-urban inhabited land, but unfortunately I could not find any way to quantify what sort of land mass that might be so no way to translate (just yet) to W/m^2.
This does not include any effect for land use (concrete, roads, structures, etc) – we are just talking about waste heat from power consumption. I think it’s also important to note that there was 300% increase in power use from 1965-2005 (5 TW to 15 TW).
This is a VERY rough/sophomoric attempt, does anyone know of anyone who has done real research from this perspective?
Correction: “in 2006 total power consumption was 15.8 TW”
That’s fascinating, but I’m wondering if you can really compare the “heat island” effect that a city has based simply on population. Would a city in a warm region have the same kind of impact on temperature readings as a cooler city?
For example, would the same method for calculating the warming bias of, lets say, a station in Lima Peru be applicable to a station in St. Petersburg Russia? Or is it possible that the warming bias is completely different?
Also, does a heavily populated area that is primarily residential and commercial have the same effect as an area that is of the same density but has more heavy industry?
You’d really have to look at a huge number of stations to make sure that these kinds of issues were not clouding things.
sphaerica (15:12:17) :
“Correlation is not causation.
Can anyone name a mechanism that would cause less densely populated areas to warm more?”
Sure, I grew up in a farming community, we had cow pastures, corn fields and broad leaf tobacco fields.
The tempuratures in the summer in the cow pastures was ‘warm’, in the corn fields ‘sweltering’ and in the broad leaf tobacco fields ‘unbearable’.
I’ve been over the description of the procedure again and again. I don’t get it. The values on the x-axis of the graphs don’t match the values you claimed you used for bins. e.g., the last value on each of the graphs has an x-axis value of about 6900. Yet no combination of the bin sizes you gave can produce a value of 6900.
I echo the sentiments of many others: you need to describe the procedure you used much better.
And the relationship between the first and the second graph? Do you REALLY use the data from the first graph to produce the data from the second graph? And if so why? Why not just directly calculate the values on the second graph.
I can see that each of the differences between bin produces a unique point on the graph; so it is conceivably possible to use the data from the first graph to produce the second graph, by virtue of the fact that any pair of bin values produces a unique value on the graph. My concern, though, is you have log-linear bins, peform a very complex operation from a statistical perspectifve on the data, and produce a log-linear output. Is it real? I can’t tell.
If your point is that there’s no relationship between temperature difference, then just say so, and move on to the correct calculation.
The correct calcuation, surely, is to calculate the warming-relative-to-zero calculation before binning the data.
Oh gee. And how DO you perform the warming-relative-to-zero calculation?! IT would be odd to use a linear projection, and then produce a result that you shouldn’t have used a linear projection, becaue the correct relationship is log-linear.
THere’s something terribly not right here. And you don’t give enough info to tell what it is.
sphaerica (15:12:17)
I guess another trend in the last 50 years, which hasn’t been discussed to my knowledge specific to climate, is a decreasing average number of people in a household (for the US at least – ref*), falling from 3.6 in 1940 to 2.57 in 2003 so for the same population we’re talking about a 30% increase in number of homes since the 40’s, but back on topic…
There are 247 acres in a km^2, so by the time you have 635 people in a km^2 (going off US numbers, not accounting for streets) you should have at least a home on every acre. Typical suburban density is probably 4 times that, but I would imagine that with a relatively low density (at least 1 house per 2 acres) you’d have a significant footprint of roads – which could probably account for a good portion of the early increase.
Can I suggest the next effort be a look at GDP (and/or other economic metrics) as proxies for UHI?
If the proposed mechanism for UHI is urban development, above basic heating, then population growth in a third world shanty town on a city’s outskirts may not deserve equal weighting with a first world apartment blocks, car parks and factories etc…
A larger economy, a higher GDP, should translate to higher levels of development and energy use, and perhaps, a better proxy for UHI.
It may also be more accurate to look at particularly energy intensive and heat retaining economic activities: Manufacturing, housing/property development, electricity use, fuel use (coal, oil, gas, et al), etc…
And considering asphalt seems to be a principle UHI culprit, what about the area of sealed roads to UHI? Or sales/use of asphalt to UHI?
A large proportion (i.e ~2/3rds) of economic activity in first world economies are in the service industries, which would be less energy intensive, although may still serve as useful proxies for levels of property development affecting UHI.
The main problem seems to be an accurate and long record of such economic data, and, perhaps more fatally, regional (as opposed to national or state) records of economic metrics, to compare to regional temperature records. But some of the metrics will be available for some regions, and it would be interesting to see if a correlation exists where a comparison can be made.
wakeupmaggy (08:27:10) : writes
“… there was a fascinating old topic with loads of photos, showing the unexpectedly high infra red camera temperatures of buildings, far surpassing the rated” tolerances of roofing materials, for instance. This was not about heat loss, it pertained to the exposed surfaces of buildings, vertical, slanted, flat etc absorbing heat from the sun. Can’t find it.”
Hi wakeupmaggy!
Try http://www.thermoguy.com/globalwarming-heatgain.html
The content has changed a bit, but the advertiser may still have the old material. It is still instructive.
Hugh
I am a nit picker. I would really like to have a look see at the micro “climate” and macro “climate” around each of the chosen sites. After Anthony’s surface station survey I think this would be a very important second pass for the data now that the stations of interest have been identified.
Is the “rural” stations sitting in a farmer’s pasture, bare ground cropland or next to someone’s barbque? Is the “rural” small town station sitting in a parking lot next to the apartment’s heat exchanger unit? Is the “city station” sited on the dairy farm sitting in the middle of the city of Rochester?
These type of questions need to be answered to make the study robust and given Anthony’s corp of volunteers the answers could be found for most of the stations.
So, to conclude we haven’t seen any explanation from dr Spencer whatsoever as to why is it more appropriate when analyzing the USA data to try to “correct” Jones or NCDC data-sets using various mathematical techniques, rather than to simply compare the rural and urban trends? There might be too little rural stations with the long record in other countries (as dr Spencer notes), but there are plenty of them in the USA.
So, dr Spencer, what do you think about dr Long’s finding that rural warming in the USA 48 was only 0.1 deg C during 20the century and 3 times lower than UAH 48 trend 1979-2009? Are we going to have any answer to that?
The argument invoked both by dr Spencer on his answers section on his blog, as well as by some commentators here, that rural stations have a larger potential warming bias is in my opinion a red herring, since we are talking about the rural stations that have never grown whatsoever, so did not have any warming bias at all. As dr Long had shown they had 6 times lower trend than urban ones.
If method proposed by dr Spencer produces just a little bit lower temperature trend for the USA than Jones finds, (and 5 times higher than the rural trend) then it is highly questionable of what use that same method of correction could be if applied to other countries? Why would anyone believe it would produce any better results there?
I know I have come a bit late to this discussion, but this analysis appears to have a serious flaw – basically a warm bias. Unless I have misunderstood, Dr Spencer uses station pairs to calculate warming per population density incease against the 2-station pair population density average. He then integrates the area under this curve to give the UHI (or station warm bias against population density – I will call this the ‘UHI plot’). If the UHI in reality has the logarithmic profile (or decreasing gradient at higher populations) expected then this process must lead to an overestimate of UHI.
Here is the problem, estimates of the warming per population density increase at a 2-station average population of 1000 (say) could include station pairs (2000,0), (1500,500),(1100,900) etc. Reading off Dr Spencer’s UHI plot gives estimates of warming per population density increase of 0.55, 0.3, 0.2degC/1000 respectively. ie reducing with reducing population spread. All will be overestimates of the true graph gradient at that point if the logarithmic shape holds.
If you don’t believe me, a simple test would be to take example station pairs from a ‘model’ with UHI-population relationship given by the Dr Spencer’s UHI plot. Run the analysis on these and see if you return to the same population density-temperature bias relationship. I would predict a bias towards overestimate of UHI. If the method cannot return the correct result from such a basic test then its usefulness must be in serious doubt.
The should be some contolling for geography. Population density correlates with proximity to oceans and seas and away from elevation. Without controlling for geography in some way, the question is left whether you have found a correlation with population density or merely that both your variables are dependent on the same true independent variable, geography.
Am probably missing something, but here goes: Let N(t) be population size in year t and [CO2]atm(t), atmospheric concentration of CO2, likewise in year t. Dr. Spencer’s analysis identifies log N(t) as a critical variable. Why not take the next step and perform stepwise multiple regression against log N(t), [CO2]atm(t)? That would tell us how much of the variance in temperature can be accounted for by log N, and, of what remains, how much is due to atmospheric carbon vs. the passage of time. Significant dependence of warming on the last could reflect several factors – among them: 1. increasing per capita heat generation resulting from greater power consumption / heat production (a likely possibility in the desert southwest as swamp coolers give way to air conditioners) 2. regional changes in land use, including greater extensiveness of individual UHIs. 3. long-term climatal changes, e.g., post-LIAA recovery, that are presently poorly understood.