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.
Ivan (21:30:46) :
Iws:
Intuitively it seems that a stable rural site with little increase or decrease in population would represent the planet’s temperature fairly accurately. Who cares if the cities are getting warmer ?…Put another way: If we are looking for the effect of CO2 why include anything except primitive sites which remain primitive ?
======
AMEN!
Double Amen.
Also, I don’t care about the small “sample size.” I would rather have one truly good site per state than a thousand crappy sites.
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?
Chris
Thanks for posting this interesting proposal. I note you have found that the most significant effects are at stations with low population density, i.e. rural stations. I have 2 questions.
1. Could it be that the term ‘UHI effect’, which was originally introduced in relation to larger cities, is no longer an appropriate term in this context? Do we now need a term such as ‘urbanisation effect’ or ‘built environment effect’?
2. By always attempting to correct for ‘UHI effect’ (or ‘built environment effect’) is there not a risk of ignoring or playing down the significance of real local anthropogenic effects on climate? Working as I do in a park near the centre of a large city, I notice what seems to be a real year-round day and night temperature difference between the park and the surrounding built-up area. The park always seems cooler. This seems to me to be a real anthropogenic climate effect, which is part of global climate. Why does it have to be corrected away? Why not just accept it as it is?
As just a bozo on the bus, I tend to agree that intuitively the best stations would be those in permanent wilderness areas. So another “amen” from me.
That said (maybe this is naive, so feel free to shoot it down), since in heavily urban areas (as long as they remain that way), there is least change in UHI over time, could they be more reliable than many rural areas that are merely assumed to have been static?
IOW, are the best stations sited in complete wilderness on the one hand, and in huge cities on the other?
Bad news. If its already been published (eg here – on-line), then a journal won’t publish it.
Journals only publish what has not been published before.
Satellite Remote Sensing of Land Use Change in Alabama of all places.
http://www.directionsmag.com/article.php?article_id=365
“The spatio-temporal information afforded by satellite data can be used to illustrate where changes have occurred and to quantify the rate of urban sprawl and concomitant loss of green space. These data provide the basis for understanding urban growth changes within an historical perspective. ”
Remote sensing by satellite of land use data is available and should be possible to correlate to historical and present temperature readings.
Both manual reading on ground and by satellites.
I suspect there is a strong correlation between land use and temperature that will explain most of AGW in CRU/GISS temp sets.
Dr Spencer,
Very interesting work.
A couple of points caught my attention. Whilst a logarithmic function might fit the UHI up to a population density of 1500 to 2000 people/sq km, beyond that your data clearly shows a simple linear y=mx+c relationship. Moreover, the slope of this linear portion is very steep in the context of historical global warming. For example, from your graph, the UHI effect of 2000 people/sq km is, by eye, about 1.1 degrees C. At 7000 people/ sq km this rises to about 1.6 degrees C, a 0.5 degree difference. In the context of a 0.6 degree C twentieth century temperature rise, this is very significant. This is an important finding and it contradicts what Jones and others have to say on UHIs, namely, that there is little additional warming in older towns and cities, even though their populations will almost invariably have increased significantly over time.
Another thought is that airport based sensors may skew the data comparison (satellite vs ground based thermometer) since a disproportionate number of land based temperatures are measured at airports since this data is needed by pilots. Even the largest airports have a zero population density, since nobody lives there. (I don’t know how SEDAC derive their population density data or how it is gridded but a large airport in a densely populated area, such as Heathrow has the potential to greatly reduce the mean population density of the grid in which it is contained)
Andy
There’s an obvious criticism of the methodology here which you have already partially acknowledged, but which I think may potentially be badly skewing your low density UHI calculation. You mention that airports have a very low population density but may be subject to a lot of concrete and jet blast. This is true. More significantly, though, a lot has changed over time; 60 years ago, a local airfield may have been a military facility, and may have had a weather station, but could quite conceivably have had a grass strip and been serving piston engine fighters and bombers. 40 years ago, many “local airports” that would have had a weather station might have had a concrete runway and a perimeter track, but would mainly see landings from DC3s carrying the mail, rather than domestic jet passenger traffic. Today, the area of apron and taxiway, and the sheer number of jet aircraft movements, combined with the length of holds at thresholds, at the vast majority of the world’s airports, has increased vastly. The character of airports has changed out of all proportion. In addition, once an airport has developed in this way, the population density around it is unlikely to ever fall for so long as it continues to operate.
All of that is speculative, but the impact (or otherwise) of population growth and/or jet traffic on airport or airfield sites specifically could be analysed in isolation by looking at the temperature records from airfields that don’t have a population driven effect. Perhaps island airfields like Diego Garcia and Ascension Island. Ascension in particular might yield some interesting results, since it has a tiny population and very limited air traffic most of the time, but became the busiest airport in the world for a period of a few months during the Falklands Conflict in 1982.
However, I’m wondering what your figures for small population increase warming effects in low population density areas would be if you excluded all airports and airfields from consideration. Would that leave enough low population density sites to be meaningful?
Thank you Dr Spencer for this interesting analysis. I would suggest some other ways to look at this.
1) Is there a way to categorize land-use or type? For example, if the nearest 80% of the 11×11 km area adjacent to a site were forested for example, it may be possible to break the effects out. Interesting areas would be urban, agricultural, shoreline, mountainous, etc. There would be difficulties in the analysis and some overlap I suppose, but different pair comparisons might have something interesting pop out. Maybe each area could be classified in a useful way and a land-use parameter assigned and cross correlations developed.
2) It would be interesting to see the same analysis applied to the 4 time of day observations before they are averaged.
3) Can the pairs also be plotted in terms of latitude (average between the two points)? This may reveal something interesting.
4) In the first graph, before the accum graphs, analyzing these separate groups may show other variables to attempt to control as you have with altitude. The variability tells me there are some factors not being controlled and possibly some weird data points with great influence. I’m confident you’re not using anything with the word “Yamal” in it, however.
5) It would be interesting to do the same analysis on, say, 1950 or 1930 data.
6) It would be interesting to see the effects on selected major urban areas (top 10), and then again on selected mid-scale cities (maybe one or two per region), then again for “best of the best” rural areas with good siting for a comparison of “accepted” temp trends versus temp trends adjusted for population growth per your analysis. How much of the accepted “signal” drops away in these areas? I suspect the analysis (since it is averaged over many pairs) might over-compensate the “best of best” stations.
Great work, and thanks again for the opportunity to comment. I’d be happy to see what I can do with the data if you’re so inclined. I suspect some of the folks at the Air Vent might have a blast with this data as well.
Ivan 19:46:47 “That’s all fine and good. There is only one problem – what is the purpose of the entire exercise? Dr Long already has shown that warming at RURAL stations in the USA is no more than 0.1 deg C during 20th century. It seems reasonable to me to assume that this rural trend represents the real climatic trend. What is the exact purpose of speculating whether the artificial warming is greater when rural area becomes small town or when the small town becomes a large city.”
Yes, a valid comment if we never needed to know about human effects on planet Earth in all their varieties. If I read correctly, Dr. Spencer’s research is a valuable statistical method to do a first “down and dirty” (Jim Evans, if I remember correctly) assessment. Dr. Spencer can amend this research as he finds helpful critiques, but get it published. Then fine-tune it with follow-up contributions by “everyone”. Let’s find out about UHI and rural (wilderness, agricultural, and land-use) changes to our environment.
Temperature seems to be a key ingredient of our future given the changes Earth has gone through in the last 100,000, 20,000, 11,000, 1,000, 350 years. Anything we can do to keep Earth warm and livable should be part of our scientific-research future. Knowledge, real knowledge, comes first, (including what it takes to maintain “warmth” during the varieties of Earth’s orbits around the sun [corrected Milankovitch] and passage through arms of the Milky Way — tiny matters, of course).
Thank you, Dr. Spencer.
Here’s an image of Urban Heat in Bologna taken by Soici Noguchi from the International Space Station.
http://twitpic.com/15huu4
He is Tweeting live from the Station
http://twitter.com/astro_soichi
Peter Sørensen (01:31:21) wrote:
“One problem I can see is that there are changes to an area that do not depend on population. A small village with 5 houses surrounded by dense forrest in 1920 could change to a small village with 5 houses surrounded by corn fields in 2010. This would have a profound influence on the measured temperature.”
Comment from Hugh Roper: your point is true enough, but that situation as I understand it no longer comes under the heading of UHI effect (or ‘built environment’ effect as I prefer to call it). It is rather a matter of changing rural land use. Changing land use is already recognised as a climate forcing by IPCC, although they seem to limit their consideration of this forcing to very restricted conditions, such as the radiative effect of snow-covered arable fields as compared to the radiative effect of previous coniferous forest at the same location. This example would be a negative forcing as it would increase the local albedo in winter. IPCC does not give much consideration to other effects of changing rural land use. However 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.
A broader and more scientific (i.e. non-missionary) approach than that offered by the IPCC to the question of radiative and non-radiative climate forcings arising from land use changes is found at the blog of Roger A. Pielke Snr:
see http://pielkeclimatesci.wordpress.com
As far as I know the IPCC ignores the extensive published and peer-reviewed work of Pielke Snr and his co-workers in this important field of climate science.
Regards,
Hugh
I’m intrigued by the slope of the bias/population density curve at low populations too.
I bet one big effect would be changes to the nighttime air inversion due to radiational cooling. Note this fits in nicely with the greatest UHI effects being at night.
Radiational cooling inversions are typically very thin, only 10 meters or so. Their formation and dissipation are quite clear on my weather station plots. While I can spot them with just temperature data, temperature and wind makes it much easier. My station doesn’t include sky cover, that’s a vital component to readiational cooling.
Adding some population and hence concrete structures to counteract radational cooling could result in big delays in forming the inversion.
Further areas of study of overnight low temps that might yield interesting results include:
Wind – if there is any wind at all there’s likely no inversion.
Terrain – the inversion forms best in broad valleys. On hillsides radiational cooling manifests itself as a cool flow of air that flows down hillsides and mount stream valleys. My wife and I have take steps to counteract this on our mountainside property in New Hampshire. On plateaus an inversion can form, but it will try to drain off a bluff. While little like a plateau, the Concord New Hampshire airport is on a plain near the Merrimack River Valley, and they have dramatic radiational cooling. It doesn’t last long in the morning.
Geology – The dry porous soil of the high deserts, e.g. around Flagstaff AZ allows dramatic cooling as the sun sets. OTOH, the high basalt walls in river valleys in eastern Oregon were not appreciated in the summer of 2003 as the basalt soaked up heat during the day and baked our camp sites at night.
Unfortunately, most of the climate data we have is merely temperature and precipitation. However, much of what I suggest here can be studied with new observations, and very possibly with http://www.wunderground.com‘s network of private weather stations. ASOS sky data should be a useful proxy for sky cover for private stations in the area.
Dr. Spencer:
Many thanks for triggering such an intriguing debate. I haven’t had time to read all the comments but clearly many have offered some interesting refinements. I recall looking at the trend data for Cambridge Bay a very small Canadian Arctic settlement and seeing a clear link between the number of dwellings and the temperature trend.
The amazing thing is you have shown that simple change in population density is strongly associated with temperature and therefore needs to be treated systematically in any assessment of global temperatures.
My guess is that Ross McKitrick will have a smile on his lips. I suspect that Roger Pielke Snr will also be very interested since I would assume that the primary thing that happens on average when a population density changes for 0 to 20 is that the new arrivals cut down the trees!!
Many thanks for such a stimulating approach to the problem. Perhaps it will be sufficient to get demographers, geographers and others to take a more systematic look.
Dr. Spencer:
One afterthought. I assume you are simply identifying a significant non-CO2 anthropogenic effect on the local temperature record? Population density is obviously a component of the aggregate UHI-effect but not the aggregate UHI effect itself.
Roy
Note Zeke finds substantial differences in the adjustments between Minimum vs Maximum temperatures. See: A detailed look at USHCN min/max temps at The Blackboard
You may find some interesting effects when you split your temperature differences with population density by time of day.
James Sexton, Wren,
Dr. Spencer is working purely with population density, not change in population density. That is explicit throughout his post. Hence, this is pretty much an argument for the opposite of a UHI, or at best, has no impact on UHI as related to population growth. If anything, it suggests (as other studies have found) that temperature anomalies are hidden by, not exaggerated by, urban areas.
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.
Leif Svalgaard, Michael Jankowski,
I would hope that an anomalous reading from a nearby fire would be recognized and corrected (10C, 10C, 10C, 20C, 10C, 10C…). Even so, I doubt that could possibly contribute to a trend seen over 10,307 data points.
The argument about UHI areas already being warmed is probably better, and adds weight to the argument that actually temperature anomalies are higher, not lower, than reported as a result of UHI.
But, again… this study needs to be done addressing changes in population density, rather than simply static population density. That is what would be interesting. Over a ten year period, how does the difference in temperature anomaly compare to the change in population density.
David L. Hagen (06:48:48) :
What I find interesting is that Carrot Eater agrees that Zeke has reproduced Mene 2010.
Zeke’s analysis clearly shows that the Data has been manipulated to raise the Overall Trend.
Enough Said.
sphaerica (07:32:21)
Clearly Dr Spencer is not measuring the population dynamics, but he has shown that as population density increases so does the temperature albeit at a declining rate. I do not see how you can possibly argue that “this is pretty much an argument for the opposite of a UHI, or at best, has no impact on UHI as related to population growth. ” What you might argue is that UHI is harder to detect from simple population numbers as the population density gets higher. However, this stems from the fact that population density captures but one component of UHI.
This study is an attempt to quantify UHI only. It does not take in to account any possible bias introduced by the very few stations currently used by GISS, or their methods of arriving at a climate trend as illustrated by current studies being done by E.M. Smith, A. Watts, and others.
I like the idea, expressed in a few posts, of taking T-min and T-max readings at various locations currently used by GISS, and spreading out from there in equal distances with a calibrated recording instrument, at for instance, two mile intervals, striving for the same altitude, equal distances from large bodies of water etc, wherever possible, progressively working toward a true rural reading, and then achieving an UHI estimate for the actual stations used by the advocates currently in charge of NASA GISS.
Re. sphaerica (07:32:30) :…”actually temperature anomalies are higher, not lower, than reported as a result of UHI.”
USA rural only data show just 0.1 deg C of warming in 20th century, and so if what you say is correct, and the rural areas actually have more UHI induced warming bias, then it is cooler then we thought.
The reality is neither warmer or cooler can be stated at this time, and the advocates at GISS decide which stations to drop and add every year, and how to transpose those measurements to nearby dropped stations, and therefore any analisis has to consider time factors relative to what GISS has done, an almost impossible task, to an opaque “science” never intended to measure a planetary heat mean.
I hold that global average surface temperatures whether heat islands or otherwise are not going to get us any closer to answering the big question. Other than producing interminable arguments about measurements, validity, analysis, and statistical methods.
Global average surface temperatures was an atempt of the current theory to find warming and relate it to CO2. It is now discredited. We must stop data mining.
If an ever increasing number of stations used by GISS, move to an area where the anomaly is increasing, new and or expanded airports, then the transposed numbers to the dropped statons will show an increased trend.
Also airports tend to be built in low flat areas surrounded by higher land, and sheltered from the wind compared to higher elevations. This kind of location, particularly in an urban setting, dramaticaly increases the UHI effect, especially at night.