Readers recall that I carried a guest post from Dr. Roy Spencer on what he learned from analysing the CRUTem3 data: Spencer shows compelling evidence of UHI in CRUTem3 data.
Now he has turned his attention to what I consider a better data set – the ISH data set, and comparing it to CRUTemp3 the results are surprising.
He writes:
The impact of making regional — rather than whole U.S. — population adjustments on the U.S. average temperature variations results in only a slight increase in the resulting temperature trend I posted yesterday, which is still well below that computed from the CRUTem3 dataset (click for high res. version):
Here are the population-adjusted temperature variations for the 3 northern U.S. sectors, with just the trailing 12-month averages plotted to reduce the messiness:
…and here are the 3 southern U.S. sectors:
The bottom line is that there is still clear evidence of an urban heat island effect on temperature trends in the U.S. surface station network. Now, I should point out that most of these are not co-op stations, but National Weather Service and FAA stations. How these results might compare to the GHCN network of stations used by NOAA for climate monitoring over the U.SA., I have no idea at this point.
Also, I need to clear up a misconception…the adjustments I perform do not remove the trends in the data. They remove only the component of the trend which is due to population density, using the regression coefficient alone (not the regression constant). There are no adjustments in January 1973 (the beginning of my data record), and then the adjustments increase linearly with time.
More at:
Regional U.S. Population Adjustments to Surface Temperatures Since 1973: Still Little Warming
On Roy’s site there is an objection to extending the line down to zero population regions, on the basis that UHI is only relevant to larger centers. As I posted there,
“Even small population centers have VHI (Village Heat Island) effects. It also depends on specific siting — conveniently close to a building, or someplace you have to put on boots to get to, every day?”
Error bands ?
@ Brian H (April 8, 2012 at 1:07 pm)
Yes, but the likelihood of having a station near to an artificial heat source (like a road or building) surely increases in proportion to the size of the population? Or am I reading this wrong?
Each building has electricity going in and heat leaking out. Even small towns have some UHI !
….someone go back and fix all the past adjustments that retroactively show the past cooler first
Back in the article is your answer:
By definition FAA stations are sitting at airports and NWS ones typically sit at meteorological offices. I believe in both of those cases you will find plenty of pavement and buildings near those sites no matter what the population is.
The problem with the Co-op sites that are used in the GHCN is that most of them require a person to physically go and read them. Human nature being what it is, not too many people are going to drive out into the sticks and hike out into a field twice a day to take readings at a station that has noting around it. Instead they place the thing at the nearest building, even in places like a National Park. An example of this is the Co-Op station that was on Assateague Island, the station was placed ON the Ranger Station.
Bottom line is that I believe you will find that site placement has little to do with population density and more to do with Human comfort and objectives.
A constant UHI offset over a period of temperature measurement is inconsequential as the argument is over temperature anomalies. I recall reading one of Roy’s posts showing that the rate of change of UHI was generally greatest in areas of small population. This is distinct from micro-siting issues such as the temperature sensor being near an AC unit. If you had a well sited instrument in a large metropolitan area, you might expect the UHI to be constant. In a village or hamlet, minor development is likely to generate a higher rate of UHI change.
David, UK says: “…but the likelihood of having a station near to an artificial heat source (like a road or building) surely increases in proportion to the size of the population? Or am I reading this wrong?”
Brian H is right. A low population density merely makes it likely that there will be lots of ideal sites for the thermometer. Lots of good sites makes no difference if the guy who has to read it doesn’t like going out into the boondocks to do so. If he possibly can, he’ll have it right close to where he lives, or, as a minimum, next to the road he uses to get to it. I don’t think there are a lot of sites with no road to them.
I have often stated that until someone actually goes out and has a look at the changes that have occured near all temperature sites over the 1970-2000 period (which was a period of quite rapid urbanisation even in small country town) then we really couldn’t have any real idea of the UHI effects on the land temp calculations. BEST’s attempts were puerile at best !!
This study seems to be a pretty good approach to what needs doing if any reliability is to be attributed to land temps..
And as suspected, there isn’t much real temperature rise at all over that period. !
The land based global temperature does have a positive trend, and it is mostly man made : UHI effects, and Hansen, Jones adjustments !!
All of this back and forth about UHI and station siting is much ado about nothing in reality.
The reality is that the vast majority of the long term land temperature records are unreplicated, non-random, worthless grab samples with far too many unknown adjustments and without any scientific / statistical utility or validity.
It is impossible to draw any useful conclusions from data with unknown variance and error. Any conclusions that are drawn from such “data” bear more resemblance to reading entrails than real science.
That is the bottom line. At least the new USCRN has some replication of measurements, “continuous” recording and more careful siting. A better start and WAY past time.
Commenting on Brian H April 8, 2012 at 1:07 pm
“On Roy’s site there is an objection to extending the line down to zero population regions, on the basis that UHI is only relevant to larger centers. ”
That would be Mosher? “UHI varies with Latitude It’s higher in the NH than in the SH”
I presume he is talking about the records – otherwise, this makes no sense 🙂
“Even small population centers have VHI (Village Heat Island) effects. It also depends on specific siting — conveniently close to a building, or someplace you have to put on boots to get to, every day?”
VHI – I like that.
I and others have studied UHI/VHI in the context of urban development around 19°S 146°E. Nothing formally documented. One source was Su San Lee, PhD Thesis, Natural Ventilation and Medium Density House Forms in the Tropics, 1998, Institute of Tropical Architecture, James Cook University.
David, UK says April 8, 2012 at 1:22 pm:
“Yes, but the likelihood of having a station near to an artificial heat source (like a road or building) surely increases in proportion to the size of the population? Or am I reading this wrong?”
The work I did was not related to measuring stations, but arose because of a concern that recent forms of urban development were counter-productive in terms of natural climate response and energy efficiency, backed up by numerous complaints from residents to elected representatives, eg “… too (beep) hot here – ya can’t go out in the back yard ” etc.
A few conclusions:
UHI can be created by relatively minor adverse dispositions of buildings and artificial surfaces to solar orientation and prevailing wind.
Long barriers of any kind (buildings, paling fences) will have a significant effect.
Natural airflow at single storey building height begins to “skim” (eg above roof height) where site coverage goes above ±29%, and is frequently absent at 40% and above.
The presence or absence of vegetation, natural or planted, can have a significant effect either way.
Not enough sites to determine if the effects were proportional to population. We were looking at configurations that had dwellings with average household size of 3.12, density around 12-15 dwellings per hectare. “Critical” sites identified varied between maybe 6 to 20 dwellings, so very much micro-climate work. The effects appeared to be related to the number of adverse configurations present, rather than population.
FWIW
“On Roy’s site there is an objection to extending the line down to zero population regions, on the basis that UHI is only relevant to larger centers. ”
That would be Mosher? “UHI varies with Latitude It’s higher in the NH than in the SH”
#####################
No that would be Zeke. Basically, Roy fit a log curve and then extrapolated down. There are huge problems with this. Lets start with the population density data.
1. It’s highly modelled and the agency that created it does not supply and PRODUCER accuracy score. If you read the papers on how it created you will come away skeptical of the accuracy of the numbers, ESPECIALLY in low population areas. Basically: people get imputed within administrative boundarys, ( > 5000 people ) and then the excess get spread out.
2. The population data is for RESIDENT population. That doesnt work so well for places like airports where AMBIENT population is more important. It also doesnt work for areas that are industrial. They will have zero population ( per the census) but lots of UHI.
However, if one wants to work with population data and sweep those concerns away, then the issue that Zeke raised was about extrapolating a 3 point curve ( log) down to zero. Lots of problems there. Firstly, log is the wrong functional form since at zero population its undefined. A sigmoidal function would be better. Other folks have used ^1/4. Oke used a log chart but he
1. moved away from his position as he studied more
2. Only looked at MAXIMUM UHI and not average UHI
3. Noted that different parts of the world give different curves
4. Added a regional wind term.
5. Had a tiny sample size of large cities
Is UHI only relevant to large cities?
1. The most extensive research has been done on large cities. The effect is a function
of physical changes to the surface.
2. Looking at Tons of satillite imagery of LST ( land surface temp) you cant see any effect
assocaited with small populations. Now, if you have a refinery that has lots of buildings
( and no official population) Then you will see a UHI, but thats because its the changes to
the surface that matter. about 80% of UHI is explained by changes to the surface.
For example; In Tokyo you see a difference between weekday UHI and weekend UHI.
So, so see 100% during the week and 80% n the weekends. Why? Because the anthropogenic portion of UHI is small compared to the changes made to the landscape.
The concrete is there 24/7 doing its work to create UHI. The portion of UHI that is due
to people actually being there? well, that varies and its smaller.
3. There is one study that looked at UHI for villages. Not very rigorous.
4. I spent a bunch of time looking for an effect between the 8000 stations in BEST that
have ZERO population and the 9000 that have between 1-15 people.
Nothing. Part of the problem is population figures are not very GOOD when you are looking
at sub 15. Not accurate. So, you find Tons of airports with zero population. you find
mining towns ( workers fly in ), you find milatary bases.
US average household electrical load is around 2 kW. Source
With an average household size of 3.12, that gives ~650 W per person, average.
1000 people per km^2 means 650 kW per km^2, or 0.65 W/m^2 of heat generation.
That’s just electricity, not counting other forms of heat generation, airflow or land usage changes.
The USAF, USMC and the USN are going to be doing some base closures over the next few years. Why not find out what bases are going to be closed, saturate the area with temperature sensors, wait for the closure, deurbanization, and see what happens to the temperature in a decades time?
Doc – are you asking climate scientists to behave like real scientists? Shame on you!
I vaguely recall a similar effect on the Antarctic weather stations.
http://wattsupwiththat.com/2009/12/13/frigid-folly-uhi-siting-issues-and-adjustments-in-antarctic-ghcn-data/
Then you will see a UHI, but thats because its the changes to
the surface that matter. about 80% of UHI is explained by changes to the surface.
For example; In Tokyo you see a difference between weekday UHI and weekend UHI.
The Tokyo urban surface changes between weekdays and weekends? I think not.
The Weekend Effect as it is called, is caused by reductions in anthropogenic aerosols and aerosol seeded clouds over the weekend. Fewer cars, fewer factories operating, etc.
Over the last 50 years we have seen large reductions in anthropogenic aerosols in developed world cities and IMO this is the primary cause of the UHI warming trend over this period. It also a major contributor to all surface warming over this period, but that is a whole other discussion.
And don’t forget shutting down the town station and depending on the airport for temperature, with airports rounding up to the nearest whole degree, providing a 1/2 degree upward bias regardless of other heat sources. Airports aren’t concerned with temperature trends. They worry about air density and lift, and rounding up provides a convenient safety factor.
http://climateaudit.org/2009/06/29/the-talking-points-memo/
USA energy use per capita peaked in 1978 and has declined by a small amount since, but in general terms energy use per capita has been flat for the last 30 years. Which makes Dr Spencer’s population density (change) a good proxy for energy use (change) per area.
http://www.google.com.au/publicdata/explore?ds=d5bncppjof8f9_&met_y=eg_use_pcap_kg_oe&idim=country:USA&dl=en&hl=en&q=usa+energy+consumption+per+capita
Following up on Q. Daniels April 8, 2012 at 4:21 pm:
” … With an average household size of 3.12, that gives ~650 W per person, average. 1000 people per km^2 means 650 kW per km^2, or 0.65 W/m^2 of heat generation….”
3.12 pp/household, average residential density 12 dwellings per ha = 37.4 ppha. City has ±180,000 population over 3734 km^2. “Urban” component difficult to define, but maybe 120,000 in 25% of the area, so call it 130 ppkm^2 max. (Large open areas, river, stormwater systems etc.)
Insolation here is often 650 W/m^2, so the urban element would be insignificant, but the surface station with the longest record is within it – at the airport.
Biggest problem here is small groupings of dwellings with a/c exhausting into small, largely enclosed, open spaces that have no natural ventilation. An interesting by-product (referring to another alarmist prediction) is that mosquitoes can hang about longer in still air, so the phenomenon may increase the risk of mozzie-born diseases.
When it’s all digital, how can we know really know what the original data are? It will be so easy to rewrite history completely. On one hand, no amount of corruption will change what Nature does. On the other hand, if the idea is to make people put chains on themselves through falsification and fear, once that goal is accomplished, it probably won’t matter if the lies are exposed.
Experimenting on Mother Gaia? Sounds blasphemous. To the stake!
If you simply compare the solidly urban data pre-adjustment to post-adjustment, there should be a DECLINE in values reflecting a UHIE correction. If there is no negative difference, then a UHIE was not done.
Top 50 urban centres, before and after: what is that like?
Hansen et al say they have correctly adjusted for UHIE. You should quickly be able to see what they did. I have no clue.
Remember that the early 1970s when this data starts was a relatively cool period compared to say the 1930s-1940s. That means that over a longer period the average temperature trend without urban heat effects is very likely downwards.
If you wish to examine UHI, you need a set of weather stations that are plausibly without UHI, to act as a baseline. I studied about 44 carefully culled candidates in Australia from the year 1972 to 2006, the former being the year of change from recording deg F to deg C.
It was not possible to find a baseline “climatic” rate of temperature change in pristine sites because of the noise, which I tried to correlate with other factors, the best correlation being longitude for T max (but not for Tmin). See graphs at base of sheet 2 of the URL below.
If you can’t construct a baseline, then you can’t look for increases caused by UHI. Simple as that. Here are the data if you wish to play with them.
http://www.geoffstuff.com/Pristine_Summary_1972_to_2006.xls
What about satellite-based lower troposphere global and regional temperature
anomaly indices? Such as UAH and RSS? Surely, those are hardly prone to
UHI effects. Global UAH and RSS lower troposphere show warming trend since
beginning of 1979, at rate of .13-.135 degree/decade.
This would indicate climate sensitivity to CO2 change roughly 2 degrees C/K
per 2x CO2, except that a periodic component having period ~60-65 years is
easily visible in HadCRUT3 and explains some of this. That appears to me to be
some combo of AMO, PDO, and a ~60 year solar cycle that may only be in
existence temporarily. Therefore, I like to see current global climate sensitivity
to CO2 change being much less than 2 degrees per 2xCO2. Some works I have
done indicate anywhere from .67 to 1.5, as opposed to 3 that IPCC favored in
AR4.
Another thing: Lower troposphere as considered by satellite data appears to
me to over-50%-consider everything below the 500 millibar level, and to give
~7-8% of peak consideration to the 300 millibar level. Increase of GHGs
increases the lapse rate in the troposphere globally, and especially in the
polar regions, mainly the Arctic. Therefore, recent-decades warming in the
Arctic at surface level should indeed be greater at surface level than according
to lower troposphere satellite-based indices.
But then again, upticking of status of AMO also warms the Arctic, and does so
more at surface level than above the 700 millibar level.
What about satellite-based lower troposphere global and regional temperature
anomaly indices? Such as UAH and RSS? Surely, those are hardly prone to
UHI effects.
It depends on what causes UHI. If I’m correct and its a atmospheric aerosol/cloud effect, then there will be a lower troposphere effect as well. You see this in the dry season data from India. Although the sign of the troposphere aerosol effect is opposite to the surface data.
It also explains the satellite surface divergence.
http://rogerpielkejr.blogspot.com.au/2009/08/evidence-that-global-temperature-trends.html
I was recently at a small airport in Northern Ontario (population under 3,000). A number of newer large buildings surrounded the older weather station around which was entirely paved. On the days we were there the snow plow (a large diesel front end loader) was parked in front of the weather station warming up before clearing the runway. The wind sock didn’t work anymore (too many buildings nearby) so they erected a new one further away from the terminal. Many of the rural airports I’ve visited in the past 10 years have been paved as part of local infrastructure upgrading. Not sure that it means anything.
DocMartyn: Closed military bases are hardly going to be “deurbanized” in our lifetimes, especially ones that are currently “urbanized”. Most bases closed in the last 40 years have been at least partially reused for industrial purposes. I would dare say none of the bases closed in the US since the 1960s have reverted to pre-construction conditions.
Robert Austin @ 2:00
Steve from Rockwood’s response above partly answers your question. If the area in question hasn’t had a population increase approaching 10 times the number, or if the infrastructure has not changed, then you’re right and there should be no real change. In practice, how many airfields or roads were tarmac or pavement 100 years ago compared with today? How many building level heat exchangers (air conditioning units) were in use 100 years ago to artificially raise the temperature immediately down-wind from the buildings?
Here in Calgary, Alberta we had a couple of inches of snow mid-week that has slowly been melting ever since. The roads and sidewalks were all clear the next day (pavement, heat absorption, albedo, etc). The snow melt progressed from these initial melt points onto grassy areas. Now all that is left with snow are areas that are sheltered from wind and direct sunlight. When I go for a walk to a large park area, I see the same trends in how the snow melts, but there is more snow around because the primary melting items (pavement etc) aren’t found there.
While Dr. Spencer’s plot of temperature vs. time presents us with data points of large number, the more pertinent question is of the number of independent statistical events that have been observed in the interval between 1973 and 2012. In climatology, the canonical duration of an event is 30 years. In Spencer’s data, there are at most one of them. At least, there are zero of them. In either case, generalizations from events as few as 0 or 1 cannot be made.
Doug Proctor says:
April 8, 2012 at 8:04 pm
If you simply compare the solidly urban data pre-adjustment to post-adjustment, there should be a DECLINE in values reflecting a UHIE correction. If there is no negative difference, then a UHIE was not done.
Top 50 urban centres, before and after: what is that like?
Hansen et al say they have correctly adjusted for UHIE. You should quickly be able to see what they did. I have no clue.
____________________________________________
Why the heck do you think Phil Jones said “The dog ate my homework” and all the Climate “Scientists” have fought tooth and nail to prevent anyone from seeing what goes on in the “black box” they use for messaging and adjusting the data.
“A Goat ate my Homework” – NIWA
Here is Hansen’s revisionist corrections on temperature.
Blink graph US temp data: http://i31.tinypic.com/2149sg0.gif
US temp raw vs adjusted: http://i31.tinypic.com/5vov3p.jpg