Anthony Watts has a post up purporting to show a very large UHI effect in Reno, Nevada.
I will just take his numbers and methods at face value, even though many questions come to mind. After all, 10oF is a big jump from outskirts to downtown, but maybe that is correct and not contaminated from engine heat. Also note this, which is specifically about Reno and shows how the UHI effect is removed from the data.
(pretty convincing, no?)
The problem is the conclusion that is at the very least strongly implied: if Urban Heat Islands are that pronounced, maybe global warming is just an artifact of that.
As this is a very common contrarian talking point, I do have a guide article about it: "Global warming is due to the Urban Heat Island effect", and the points are all covered there.
I only want to reemphasize that UHI is real, no one disputes that! So it is accounted for in the reconstructions. I also want to present again something that strikes me as rather irrefutable evidence that a global warming signal is not the result of primarily urban readings. The evidence is in the images below and the plain-as-night-and-day fact that there is no correlation between warming regions of the globe and urbanization. If the "it's just UHI" folks were correct, then the global hotspots would by and large be over areas with lots of development.
Here is a global anomaly map for 2000-2006 with 1951-1980 baseline:
and here is a satellite image of global urbanization:
Note the complete lack of correlation for Canada, China, USA, India, Asia, Eastern Europe, Africa and the poles. Also note that the one place with strong correlation between urban centers and warming, western Europe, is also the place where urbanization has changed the least, the cities are hundreds of years old and have developed much less over the last 50 than North and South America.
So. Where's the beef?
So help me understand this. Are you saying that the Reno temps were adjusted to based on 10 nearby stations? How do we know that there has been no UHI or poor siteing location on any of the 10 "control" sites?
I am not intimate with the details, but descriptions of the methodologies used say that the control sites are rural. I have no doubt that there are siting problems in many places, there are after all thousands of stations and they were not intended by design for determining trends. But there would need to be a consistency in the errors that is not plausible by chance to be all in the direction of warming if you want to justify dismissing the observed trend as artificial. There will always be some degree of error and uncertainty.
There are also many other indicators of warming that support the integrity of the station data and its processing.
One of the unexplained elements of the instrument temperature record is why it doesn't correlate with temperature proxy data in the last 50 years or so. The last .4C of rise that the instrument record shows is simply absent from the proxy reconstructions. So which is wrong. In regards to the instrument records, a small town can be considered rural and therefore be used to adjust for UHI. But the problem with this philosophy is that even in a small town, someone can build a blacktop parking lot right next to your measuring station, and the result will be a rise in temp that will be used to adjust all of the urban areas nearby. Satellite data should not suffer from UHI effect. And for the first 20 years both HadCrut3 and GISS trended along well with the satellite data. HadCrut3 has continued to trend with it fairly well. But for the last ten years, GISS data has diverged strongly from it. It would be very interesting to see how the adjustment algorithms used by the Hadley Center differ from those used by Hansen. Hansen was finally egged into realeasing his code. I don't believe that Hadley has yet released theirs.
The issue with the Reno UHI is that it looks like it is both larger than it used to be and larger than the correction made to it by GISS etc.
Over at my blog (http://www.di2.nu/200810/29.htm ) I've displayed two graphs. One is a very simple one showing the enormous growth in population of Reno in the last 40+ years. The second (taken from another commenter's use of GISS data in this graphing tool - http://www.appinsys.com/GlobalWarming/climate.aspx ) shows how Reno's temperatures have increased notably compared to surrounding more rural stations. The correlation between Reno's temperature and those of its neighbours is striking except that Reno's temperature gradually increases over the last couple of decades
Temperature trends are similar over distances as large as 1000 km. I don't have details for Reno, but the ten 10 'rural' stations used to 'correct' Reno data are itself corrected in turn by other stations, so the global picture exhibits the correct trends.
As for the 'recent' discrepancy between GISS and HadCrut3, there are I think at least plausible explanations:
1. HadCrut3 temperatures are averaged over 10-year intervals, but no such interval exists for the last 5 years, hence those values are created 'artificially' or with other averaging methods.
2. More plausibly, the two methods do not measure the same things. HadCrut3 excludes polar data (as no permanent stations exist upward of around 70° latitude), while GISS does include polar data. The picture shows that a lot of warming is going on in the Arctic. This is captured by GISS, but not by HadCrut3.
"I don't have details for Reno, but the ten 10 'rural' stations used to 'correct' Reno data are itself corrected in turn by other stations, so the global picture exhibits the correct trends."
As long as any of the rural stations, or any of the rural stations used to correct rural stations have had things done, (like parking lots being built next to the measuring stations) then the error will be propogated to all of the urban stations that use that rural station, as well as any other rural stations that use it for correction.
"1. HadCrut3 temperatures are averaged over 10-year intervals, but no such interval exists for the last 5 years, hence those values are created 'artificially' or with other averaging methods."
This is certainly not true. The monthly magnitude of HadCrut3 movements are as large as any other. If it were averaged, it would look much, much smoother.
"2. More plausibly, the two methods do not measure the same things. HadCrut3 excludes polar data (as no permanent stations exist upward of around 70° latitude), while GISS does include polar data. The picture shows that a lot of warming is going on in the Arctic."
This is also not true. If warming were going on, then the extrapolation and addition of Arctic data would make that warming go faster in GISS data. But since there is no warming going on at the lower lattitudes, then there is nothing to be multiplied and extrapolated to the Arctic. Now if Hansen is increasing Arctic temperatures at a time when there is no incease outside of the Arctic, then he is corrupting his data. Let's also remember that real station data from the Antarctic shows that it is cooling.
Yes, the image is beautiful. The story behind it is ugly.
HadCRUT, GISTEMP, and RSS track very closely. UAH is the outlier.
Lets not lose sight of the viral meme which skeptics faciliate the propagation of:
Viral meme: "Most, if not all the recent warming is caused by urban heat island and microsite issues"
They don't often straight out say it, preferring to hide behind a wall of ambiguity. But they certainly like laypeople to get infected by it and go out of their way to imply it is likely.
In fact it's codswallop, there is a whole stack of evidence against the claim that most, let alone all, of the recent warming is UHI or microsite bias and deep down surely many of the skeptics must know it.
Tilo says it all: "Satellite data should not suffer from UHI effect. And for the first 20 years both HadCrut3 and GISS trended along well with the satellite data"
Exactly. So during that 20 year warming period there is strong evidence against any of that significant warming being due to UHI or microsite biases.
Of course it's more like 30 years. Appealing to a slight divergance in the GISS record over the past 10 years compared to HadCRUT hardly justifies the payload of the viral meme.
"HadCRUT, GISTEMP, and RSS track very closely. UAH is the outlier."
This is not true. GISTEMP used to track well with the others. But over the last decade it has diverged more and more from the others. Now HadCrut, RSS, and UAH are tracking very well with each other and GISTEMP is an outlier.
"And for the first 20 years both HadCrut3 and GISS trended along well with the satellite data"
Again, this is untrue and I did not say it. See above. Here is a chart showing HadCrut3, RSS, UAH and GISS for the last 11 years. Clearly GISS is now strongly divergent.
Correction to above post. I read Makron's remark as "for the last 20 years", instead of "for the first 20 years." It's true that the data for all did track well for the first 20 years that we had satellite data. What is also worth noting is that Hansen was getting the warming that he wanted for that time period. After the trend flattened out and the warming was no longer there, Hansen failed to track with the others. Coincidence? I doubt it. Furthermore, Hansen's divergence for the last 11 years is not minor. If you look at the fact that there is absolutely no warming for RSS, UAH, and HadCrut3 - that in fact they show slight cooling, then Hansen's divergence is actually half of the total warming that the IPCC says we should expect from additional CO2 for the time period. In other words, divergence alone accounts for half of the IPCC prediction in Hansen's case.
Over the entire length of the satellite record, RSS, HadCRUT and GISTEMP show virtually identical rates of warming. UAH is the outlier. These are mathematical facts. One needs only look at the data.
Or if that's too coarse for you, you can compare GISTEMP and RSS' monthly anomalies directly.
The new RSS analysis (v3.2, which I haven't graphed yet) shows an even closer rate of warming over the entire dataset.
Looking at a subset of that record to make the case that GISTEMP is the "outlier" is either dishonest or incompetent. You can't stop the clock at the height of the 1998 El Nino, when GISTEMP shows the lowest anomalies, and then restart it so as to create the illusion that GISTEMP is "strongly divergent" when any glance at the data shows that it is not. The same GISTEMP algorithm is used over the entire time period. There is nothing different about it during the last 11 years versus the 11 years before that.
GISTEMP and HadCRUT use different SST data (70% of the globe right there) different homogenization methods, different corrections for urban warming, and different interpolation methods. RSS, of course, is a synthetic MSU "channel" and is thus completely different. Yet despite all of these differences in data and methodology, the long term trends between all of them are virtually identical, although you can certainly cherry pick any number of shorter time periods to invent "divergence" between them.
Reno perhaps is not the best location to be used given that when you go west, your elevation increases quickly in a short period of time. Hence several miles out you are significantly higher and hence one would expect lower temperatures.
We live about two miles west of the city limits of Calgary Alberta Canada. I have my own mini weather station as my wife likes to record data for her gardening etc. What we have constantly observed is our temperatures at night tend to be significantly lower than those recorded in Calgary (3 to 6 degrees C). While the daytime highs tend to be the same for us and the city, it is the night time temperatures that exhibit the delta leading one to conclude that at least here the urban heat phenomenon exists and has done so for the last 20 years (we first moved here in 2010).
Now as for the posts of temperature contours exhibiting annual global data, you have to take what they project with a grain of salt. For some unknown reason the GISS data uses very little data from the artic (contoured on 1200 km radius) despite the fact that there is ample data available. The second issue with the GISS data is that it does not include ocean air temperatures, only land temperatures which will skewer the results to higher levels. With today's shipping, there is a bevey of data available for ocean air temperatures so excluding them from an analysis is incorrect (unless you want to project a pre-specified observation).
Another seemingly suspective manipulation involves the plotting of temperatu anomalies instead of outright temperatures. In this way the maps showing the greates anomaly tends to be contoured red (means "HOT") so when one looks at the global map of the GISS data, they observe a band of deep red across the north. Given that for the most part the artic temperatures are well below freezing during the winter and barely above freezing during the summer, an increase in average temperaturs of several degrees would not have an effect on melting in the artic (avergae moving from -20 to -15). Hence what we are seeing happening in the artic must be related to something other than the accumulation of atmospshere CO2. - my guess is the sun's radiation combined with the earth's reducing magnetic field - and not to forget the role of a warming ocean (from increased radiation) and altered air currents.
Here are two site that elaborate more:
For those people who continue to believe that Urban Heat Islands are a myth should boggle the mind of any pragmatic reasonable scientist.
"For some unknown reason the GISS data uses very little data from the artic (contoured on 1200 km radius) despite the fact that there is ample data available."
No, this is untrue. Please try to substantiate this with a link to ample temperature data covering the arctic over the last 100 years and get back to us. There is so little data that the HADCRU global temp reconstruction does not even try to include the arctic. The fact that GISS tries to do so using interpolation methods and the scant data there is, is the primary reason their reconstruction shows slightly greater warming than HADCRU. The arctic is warming much more than the rest of the world, leaving it out of the global average skews the resulting trend downwards.
"The second issue with the GISS data is that it does not include ocean air temperatures, only land temperatures which will skewer the results to higher levels."
I see. Then how do you explain this: "Line plot of global mean land-ocean temperature index, 1880 to present, with the base period 1951-1980.", linked from this page. There are also links to land only, NH land-ocean, SH land-ocean and mid lattitude plots as well.
That makes two complete misses on simple matters of easily verifiable fact. So when you get to: "Another seemingly suspective manipulation..." it is kind of hard to expect anything of substance. Your "suspicion" of the choice to plot anomalies does not disappoint and you seem to be unaware that summer temperatures in the arctic do get above freezing. Check this beautiful picture from Greenland and don't forget the annual retreat of sea ice.
Finally, you close with: "For those people who continue to believe that Urban Heat Islands are a myth...", which is a nice, in a nutshell summary of the whole UHI denialist talking point. "Those people" do not exist. From the original article above: "I only want to reemphasize that UHI is real, no one disputes that!". The UHI effect is real, it does not however expalian away the rising global average and it is discussed and accounted for climatology. So how about arguing against a point that was actually made, like the complete lack of correlation between urbanization and warming? I have yet to hear a peep from anyone on that point.
Geo-guy, given the fact that you got very basic things about temperature data just plain wrong, and I have demonstrated it clearly, would you consider the possiblity that the sources you posted for us are not in fact reliable? Do facts matter to you?