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.
“ColorMeSceptical (15:23:14) :
I’ve been lurking here for a long time, often impressed by the quality of work from Dr Spencer et al. But this time I hear alarm bells ringing: Can a population of 20 per sq. km really raise the temperature by almost 0.8 degrees C? [….] ”
I have just done some preliminary daytime measurements with a datalogger and I have found that just passing through a road construction zone in the country is enough to raise the temp by nearly half a degree C, and that a small hamlet can have a similar effect.
son of mulder (15:27:59) :
” Ivan (12:51:10) :
I really, really cannot understand what is reason for all of these mathematical speculations. Is not far easier to select ALL RURAL stations in the USA and compare thus obtained trend with the urban trend? Is that idea really so unimaginable and stupid?”
Keep asking the question Ivan. Maybe someone who has the raw data will do this. What is there to lose? Let’s us just be told the rural trend.
I’m currently trusting no one. But the truth is still out there! Give us the data. Hypotheses can follow.
=====
It would easy to have a an urban series and a rural series. But I’m not so keen on raw station data. If a station has crazy results, why include it? I suppose the stations that overstate temperature could offset those that understate, but I wouldn’t just assume it.
Well Roy, I didn’t anticipate exactly what you were trying to achieve from your previous posts. Well done for the ‘secrecy factor’, I concur that it’s better to keep the ‘rest of the field’ guessing when a hypothesis is in the embryonic stage!
Then again, why submit a hypothesis that supports a regime that can’t show climactic resolution without masses of ‘noise’? At the risk of a post edit I’ll include a post that I made on Tamino’s site to J (which was also edited):
J.
“If you think you’ve found an error in Tamino’s calculations, or a logical flaw in the conceptual framework for those calculations, you need to explain that. “I just feel that the data are insufficient” isn’t an explanation. Or at least it’s not an explanation that anyone else will find convincing.”
OK! One last try! I don’t see that it should be the responsibility of a mediocre engineer like myself to explain a definitive signal resolution to science types here. Especially when Tamino keeps telling me that “I don’t have a clue”.
I’ll start with an explanation of a ‘resolved signal’. To get some background on this I’ll begin with a single wire telecom digital data transmission (DTX). Single wire data transmission is commonly achieved in analogue (ATX) by use of a ‘carrier frequency’. The higher the carrier frequency, the greater the definition to the analogue signal (as can be seen on any oscilloscope).
Short explanation; a carrier signal is used for ATX because different wavelengths of an analogue signal travel at differing speed vectors through different media and the practise of modulating the analogue signal onto a carrier wave (amplitude modulation [AM]) maintains an identical transmission speed vector for all analogue wavelengths.
With DTX, if we can control the digital 0, or 1, signal by a co-ordinated time control mechanism, like within a computer, that would be OK. However, between computers (especially distant network, fax and Internet, etc) a DTX needs a carrier signal to maintain the signal integrity. In short, the digital 0, or1, has to be an AM DTX. Thus a ‘modem’ (mod [ulator] dem [odulator]) of some sort is needed to integrate, or interpret, the data within the carrier wave.
Now that we have a modem we can receive data from ‘everywhere’, but how fast?
A ‘received data transmission’ (DRX) has the same constraint as a ‘transmitted data transmission’ (DTX), except that some network and Internet ‘protocols’ use verification algorithms, etc that use up the DTX/DRX stream, but we’re more interested in signal integrity here.
The transmission rate of data is commonly understood as ‘the Baud rate’ and is stated as the ‘bits per second’ of digital data transfer. A ‘bit’ is simply a ‘bi [nary digi]t’, a ‘0’, or a ‘1’, but what determines the maximum rate of transfer for this? Its the carrier wavelength, or carrier frequency to add a time constant (‘frequency’ is the inverse of ‘length [distance] travelled/second’). I’ll admit that I’m happier with frequency units rather than wavelengths and Baude rate is a frequency measure anyhow. I’ll leave this up to you to verify, but the carrier wave needs to be a maximum of 1/3 the wavelength of the Baud rate, or at least three times the signal frequency (baud rate) to qualify as validly recognisable data!
Enough of point data validification. I’m confident that Tamino is correct in the respect of analysis, but four data readings in 24 hrs isn’t a full resolution of that station’s temperature data that was analysed. Were clouds evident, did it rain, etc? There isn’t enough station data to provide a feasible data-set of average station temps for an analysis in the first instance IMHO.
However, on the point of a region being verified from a few network nodes for whatever purpose,??? I’m truly sceptic. We need to discuss and elucidate this assumed proposition.
When I was younger I used to ride a motorcycle. In summer I would ride wherever I could in the UK. The thing that struck me was that in summer you could ride and feel the temperatures alter almost every 1/3 km of the journey. It was exhilarating, but during the autumn, winter and spring in the UK it was too cold to notice much difference (a ‘normal human condition’ to local temps, I guess). Now that I use a car I don’t feel this (don’t think I’m missing much). However, this says a lot about the points of data collection and the accuracy of a grid square to represent the regional temperature.
With all due respect J, we should believe the thermometers and ask why the disbelief in them for that region (assuming the thermometers are stationary, as in the suburban/urban/rural categorisation and accurate). I like raw data. Then I can decide for myself if there’s a weighting that needs to be made. If UHI expands, it’ll show in the raw data signal. Moreover, the appearance of UHI in the raw data would validate AGW. The main problem with this is that stations (network nodes) are so far apart that they can’t define grid box definitions, such as temperature, when they are so distant. There are spatial losses that can be assumed, but not known.
For example, Box 5 has 118 stations. Randomly select 59 of these stations and calculate the box temperature. It’ll be different to the result for 118 stations! Now take the 59 stations that were ‘not selected’ and calculate the box temperature. It’s different again, but opposite. However, average the results from the stations that were selected and the stations that were not selected and you’ll arrive at something closer to the 118 stations result. What’s more it’s possible to repeat this check with as many random samples as you like to arrive at a conclusion for error bands for the ‘sample rate’ (total number of stations). Warning! Before attempting any of this, ensure that raw station data is used! Otherwise an impaired verification of ‘nodal signal : grid box resolution’ may result.
When observing the convolutions of a point phenomenon of nature its important to observe the convolutions at adjacent points elsewhere as well. A station’s data (whatever baud rate) is only a point in a network that can report ~metres from its location. We need many more land stations to resolve the definition of land temperature for a large area IMHO. Don’t take my word for it though because ‘I don’t have a clue’.
Best regards, suricat.
End of post.
Surely any money introduced into a climate budget would be better spent on the number of surface sites available and the data that flows from them! Without this caveat climate science just isn’t a science, it’s only a conjecture (anecdotal).
Before you say it, yes! Satellite data has greater resolution than surface station data, but it lacks confidence and still doesn’t have a good resolution. However, surface station data currently relies on an unchanging quality that can’t be adequately defended due to lack of knowledge for the forcings to a network node (site) that doesn’t provide coverage for the ‘region’ greater than a few hundred metres, with regard to temperature anyhow.
Best of luck with your project, but I still have doubts for its outcome. Though, I hope my post here helps rather than hinders.
Best regards, suricat.
Is there a site, near one of the new generation of proposed nuclear power stations that is virgin? Could they not set up a set of temperature stations 1/2, 1, 2 and 4 miles in rings around the proposed site, take temperature measurements for the 4-6 years before construction begins and then look at the 3-10 years during and after construction?
rural stuff
irrigation systems. they cover large areas and turn on intermittant
summer fallow. farmers may plow these under several times over course of summer.
fertilizer. depending on type may affect large areas
herbicides, same
erection of hog and poultry barns. these don’t look very large but they generate an enormous amount of effluent. how they dispose of it can affect a large area.
flax stubble. you would be shocked how many farmers burn their flax stubble
sun flowers. these are not harvested until after freeze up. completely different heat profile the next year when the farmer rotates crops and plants potatoes which are harvested before frost.
septic fields. most farms have one but in a lot of denser area there’s sewer lines going in. septic fields are pretty hot and switching to a sewer line eliminated the septic field and transfers the heat elsewhere
daughter factor. when I was young there was a farmer in the area who had 11 daughters. temperature was very warm at his farm, mostly after dark. discharge of light gauge shotgun loaded with salt seemed to cool it off.
might think of a coupe more…
more rural stuff
snow fences. farmers may erect very long snow fences to retain snow on their fields. The common explanation is that this is to raise spring moisture levels in the soil but the real reason is to influence snow depth and start fights between Willis and Steve.
So, GISS have been measuring increasing urbanisation, not global warming.
My hunch is that with time, rural stations have deteriorated, maybe more so that urban stations. I would use a rating scale having to do with quality (such as the list provided by Anthony), not a population density measure, to determine spurious heat island and paired sensors. In fact Urban Heat Island is a misapplied label in my opinion. In essence, we are talking about spurious heat. This probably has less to do with Urban versus Rural, than with microclimate changes around the sensor wherever they happen to be. The new term should be defined as spurious heat island affect secondary to microclimate changes, not urban versus rural.
Here is one reason why I think this is important. When you publish, you will be listing your stations. Your critics could find all kinds of problems with your stations that will bring doubt to your results. Unless you know the quality of your stations, I wouldn’t use them.
Side note: It amazes me how much of the world is not particularly liked as a place to live by most people, yet these very same place are where I WANT to live. Odd.
lws (17:11:44) :
Refreshing.
I have read so many “studies” with an agenda of proving something like “how much damage AGW will cause”. I am sure there are anti AGW papers with an agenda too.
I want to see science where the agenda is TRUTH ! What a novel idea !
Accurately adjusting for UHI is important and let the facts be what they are.
A great man once said
“You have a right to your own opinion you do not have a right to your own facts”
BTW : How about all of those lawn sprinklers in Phoenix. ? More people more sprinklers ? More car washes ?
For that matter how about western Nebraska agriculture sprinklers ?
Water vapor is a great greenhouse gas !
======
I thought it was already a part of the cycle, but if I’m wrong droughts could be solved by people urinating more. I’ll drink to that !
Nick Yates (17:48:07) :
So, GISS have been measuring increasing urbanisation, not global warming.
====
I guess UAH has too, although I’m not sure how.
So what? It’s anthropogenic global warming, anyway you look it at.
transcript is up…this is the opening q&a for Phil, where Phil’s ‘most’ means ‘all’….
UNCORRECTED TRANSCRIPT OF ORAL EVIDENCE
SCIENCE AND TECHNOLOGY SUB-COMMITTEE
THE DISCLOSURE OF CLIMATE DATA FROM THE CLIMATIC RESEARCH UNIT AT THE UNIVERSITY OF EAST ANGLIA
Q78 Ian Stewart…Professor Jones, there has been some speculation that the primary data has been lost and manipulated. Are all the raw data used in your various analyses accessible and verifiable?
Professor Jones: The simple answer is yes, most of the same basic data are available in the United States in something called the Global Historical Climatology Network.
http://www.publications.parliament.uk/pa/cm200910/cmselect/cmsctech/uc387-i/uc38702.htm
3 March: Quadrant Mag Australia: ABC gags Bob Carter
by Michael Connor
Quadrant Online previously reported that the ABC had invited Bob Carter to contribute to an online debate on The Drum following their publication of a series of five articles by Clive Hamilton.
Left internet newsletters and blog sites were outraged that sceptics were to be allowed to comment on their ABC.
Professor Carter submitted his article, on James Hansen and the Hansenism cult, and the ABC has rejected his article – which Quadrant Online is privileged to publish.
James Hansen is visiting Australia. We can only guess at the pressures which have been exerted on the ABC to close down criticism of Hansen – and the cowardice which saw them conform. So much for Australia’s brave freedom fighters of the press.
Read the essay the Left tried to ban, hear a voice the Left wants to silence:
“Lysenkoism and James Hansen” by Bob Carter here…
http://www.quadrant.org.au/blogs/doomed-planet/2010/03/abc-gags-bob-carter
After a bit of thought: If you still want to compare population density, I would tighten the variables so that you are comparing apples to apples. Apples grown in the country versus apples grown in the city. In other words, limit, as much as you can, contaminating variables that can later be used to bring doubt to your subject pool.
First, I would choose only stations that originally met a pre-determine set of criteria for station citing. They all started out the same.
Second I would pair similarly current categorized urban and rural stations.
The idea is that you tighten the other variables so that you are assuring that you are only measuring similarly “categorized for quality” stations between rural and urban settings.
If you don’t control for variables, you have to list them all and then deal with the extremely high required number of subjects to minimize the variables. It is a standard in research design. The more variables you have, the more subjects you must use. And the number rises exponentially with each additional variable.
geo (15:56:02) :
Can somebody explain to us if Dr. Spencer’s work here supports or contradicts Dr. Long’s recent SPPI paper on rural UHI? Intuitively my first reaction is they can’t coexist in the same reality, but my inuits have been wrong before!
===========
I had the same thought, but I haven’t checked Dr. Long’s article, so I’m not sure.
It seems like many have misunderstood what Dr. Spencer is doing. All he is doing is comparing the temperature difference between two, relatively close locations with different populations and finding that the larger the population, the warmer the temperature. Furthermore, the largest differences, on average, occur when comparing ‘near zero’ population areas with those areas in the next higher population bins. This does not imply that warming over time will be greatest in the rural areas, as some have speculated. If the rural areas have experienced growth, the warming will be much greater than the same amount of growth in an urban area, but the opposite is also true. Rural areas loosing population would, most likely, experience a cooling, provided the infrastructure around the site gradually returns to a more natural state.
This comparison of site pairs does not determine the UHI for either location. That can only be determined when you look at population changes over the decades. Dr. Spencer is just trying to calculate realistic numbers to be applied to locations with population changes. If there is no population change, then no correction is needed to the raw data. Historically rural stations are often considered the best stations to use to determine climatic changes because population and land use changes are generally minimal or nonexistent.
Is this the best way to do it? No. The best way would be to examine each site and determine all changes that have ever taken place and when. Then we would find that it is not just population affecting the readings but everything from tree growth, to paint, and surface changes and buildings and exhausts and numerous other heat sources. We would have to do it for every station and we would probably die of old age before we got half way through, but the end result would probably not be much different than Dr. Spencer’s down and dirty method.
Dr. Spencer’s method appears reasonable and, equally importantly, doable. It gives us a number to adjust the temperature of any area with a changing population over time; a variable that is generally well known. Will it be perfect. No, but on average it will be valuable. It certainly appears more ‘robust’ than the now discredited Jones et al study that was done back around 1990. That study had ridiculously low values for UHI adjustments, but was used in determining ‘global warming’ for the last 20 years.
The real benefit of the Spencer method over the Jones method is the amount of data in each. Even if Dr. Spencer has some questionable pairings in China, they will be unimportant when factored in with the other 9,800 pairings around the world. Sadly, the limited study done by Jones et al was not so immune.
Thank you, Dr. Spencer.
My very first lesson in graduate level statistics is to ask a statistician how many subjects I needed given this many variable. And that didn’t mean the number of variables I was studying, it meant the number of variables inherent in the subject pool.
A simple example: If I wanted to study college behavior, I had to control for or actually study the effects of age, marital status, living choices (IE frat house, dorm, GDI, etc), etc. If I couldn’t control the variable, it had to be part of the study, which required an exponentially larger subject pool.
Just a note: It has been more than a DECADE since I took my statistics class. So check with a more knowledgeable person. I could easily be wrong.
By the way, I am still home sick with a roaring sinus infection. So please pardon my drugged up typing and lack of attention to grammar.
more rural stuff
bird migrations. geese in particular will adjust their migration patters to avoid cities over a certain size. if this results in them choosing a different lake as a stop over point, you may get a big change in the lake. for example, lakes surrounded by pine trees are highly acidic, but goose guano neutralizes it every fall… until the birds go to a different lake. then the first lake dies off due to lackaguano disease. sometimes misdiagnosed as acid rain.
Roy
When applying such UHI per capita correlations over time, it may help to adjust for changes in energy use per capita over time. See my note above. e.g., See:
Historically, countries rapidly increase per capita energy use during a industrialization phase, then settle into lower per capita energy use growth rate. e.g. See :
Exponential growth, energetic Hubbert cycles, and the advancement of technology, Archives of Mining, Polish Academy of Sciences, Tad Patzek
http://petroleum.berkeley.edu/papers/patzek/ArchivesofMiningPAS.pdf
Note especially:
For data on per capita energy use, see:
International Total Primary Energy Consumption and Energy Intensity
US Energy Information Administration
http://www.eia.doe.gov/emeu/international/energyconsumption.html
e.g. for 2000 see:
All Countries 1980-2006 for the International Energy Annual 2006
http://www.eia.doe.gov/pub/international/iealf/tablee1.xls
Pamela Gray,
I bet that if you take a hundred climate change fanatics and off-topic presented them with two job offers, one in New York and one in L.A., the unimaginable difference in the two climates comes 33rd in their decision making, right behind closet space and nearby playground locations.
If people were serious about their climates for actual reasons, as opposed to poseurs, they’d be afraid of moving from New York to New Jersey becaue they don’t think they could adapt to the weather there.
I believe the stronger proportional effect at low densities is simply due to picking such a small grid resolution for population density. I would like to see the same analysis using the population in a 10KM grid. A lot of the low density stations are getting the heat blown in from the nearby city depending on the wind direction.
Also, a few facts:
An average person emits 60Watts. An average household consumes 1000Watts of electrical power. An average 1,000,000 person metropolitan area will consume about 1000 MegaWatts of electrical power and a similar amount of heat for transportation and direct heating. The power plants that supply the energy for that metropolitan area, typically 30-100 miles away, will emit a similar amount of energy as the whole metropolitan area.
So at 1000 people per square mile, the extra anthropogenic heat is several Watts/square meter depending on where the power is generated and how much industry is around. This explains most of the UHI without resorting to albedo changes and thermal mass changes.
I only bring up the variable contamination to prevent the “tree ring” problem, IE a subset of subjects causes most of the change.
Pamela Gray,
My kitty and bunnies send you hugs! Eat lots of chicken noodle soup, but avoid products made with bunny rabbits.
(They made me say that.)
For chronic sinus infections doctors are filling the sinuses with some kind of foam, which somewhat implies that you can snort “Great Stuff” to good effect, but I would avoid it lest your face explodes as the polyurethane foam expands.
My advice may explain why kids rely on Dr. Mom instead of Dr. Dad.
Are you sure you are wise to post that here first? Aren’t some journals sniffy about wanting to present data that is original – i.e. has not laready been put into the public domain? By putting it here first aren’t you ruining the “scoop factor” and making it less likely that it will be picked up in a properly peer-reviewed journal?
You noted that population density was not the only significant determinent using the example of airport siting. The increasing use of energy per capita may be one also, ie the UHI effect vs population for 1950 may be different from that of 2000 such that it may not be accurate to use the 2000 data to adjust land temperatures 50 to 100 years earlier.