The Urban Heat Island effect on temperature records is real, despite what some people wish you to believe. Peter, a sixth grader, and his dad, thought so too, and take the data from NASA GISS and show in a simple video, what we’ve been saying for years here at WUWT. Urbanization, land use, and station siting matter.
Watch Peter’s excellent video below:
They used a simple pairing of rural and urban sites to show the differences. This shows why homogenization, which smears all the data from urban and rural sites together, is a bad idea, and gives trends that don’t exist in reality.
I like the ending where he says in the rolling credits “Peter’s dad is not employed or funded by any energy or oil companies”. It’s funny that they’d feel a need to say this. No National Science Foundation funding needed either.
This video appeared in comments on WUWT, if anybody knows how to contact Peter or his dad, please advise. We are in touch now.
One wonders what the response of the well funded Hadley Centre, Met Office scientist Dr. David Parker, might be to this video.
Parker’s 2006 paper published in the Journal of Climate titled: “A Demonstration that Large scale warming is not urban” claims:
The analysis of Tmin demonstrates that neither urbanization nor other local instrumental or thermal effects
have systematically exaggerated the observed global warming trends in Tmin. The robustness of the analysis to the criterion for “calm” implies that the estimated overall trends are insensitive to boundary layer structure and small-scale advection, and to siting, instrumentation, and observing practices that increasingly influence temperatures as winds become lighter. Furthermore, even at windy sites (e.g., St. Paul, Aleutian Islands, in Fig. C1), the calmest terce and especially the calmest decile will be strongly affected by occasions with very light winds in passing ridges or blocking anticyclones, and should reveal any urban warming influence.
…the results of the present study also suggest that they have not affected the estimates of temperature trends.
Steve McIntyre gave Parker’s paper a scathing review in 2007’s article:
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Peter’s dad here. Hi, interesting discussion. I only checked one pair of sites last night and I’ll let Anthony discuss it more if he wishes. The data I downloaded from GISS is now different from the data I downloaded in early 2008. The rural site in the pair now has different numbers (and even different years represented) from the original download but the urban site numbers are the same. I don’t know how this will affect the overall calculations but there are now differences in the database. I really have too much work at the moment to check all the sites but I will try sometime. Anyway, to argue over slight differences in trendline equations when the r2 values are so low if really counterproductive. The variation is very high and you must judge accordingly. Quite frankly, you can’t get any accuracy unless the r2 is above 0.9. At this point you really need to think of this as a qualitative analysis not a quantitative analysis.
By the way, Peter is thrilled with the response all of you have given to his video and is planning more. Thank you all.
I will be very busy and won’t be able to comment for a while. In case you want to know, I’m checking different extraction techniques that allow me to very accurately quantitate the amount of particular micro-organism DNA in very small samples. I can actually count the number of viral particles in microliter volumes so that’s why I know more than a little bit about analyzing data.
REPLY: He’s correct, and it is what I’ve been alluding to in comments above – the GISS data has changed, significantly since then. I have a detailed post planned on this – stay tuned. – Anthony
I have to admit that I’m surprised by the friendly turn in the conversation. I turned on my computer this morning expecting a hail of abuse. Thanks. to those who are backing me up a little bit.
For those who are angry at me, please consider something: How did my aggressive comments and quick analysis compare to accusations of incompetency, dishonest, and even fraud that are commonly applied to climate scientists in this forum?
—
Bill Illis:
I’m not going to jump through any more hoops for you. You accused me of not doing the work that I did (at lucia’s) — you essentially said I was a fraud. I demonstrated that I did do the work. I put in the time to test something that didn’t look right. Anthony and Peter’s Dad have the spreadsheet (and Anthony can post it online if he wants). We will see if my results are correct.
You can re-gain some credibility with me by retracting.
gjg:
Thanks for the response. Did you have a chance to compare any of my data against the new GHCN values?
Anthony:
I have an archive of the complete GHCN dataset from Sept 13 2007 if you are interested. If the data did change this might help determine when it happened. Wouldn’t it be ironic if my PITA comments lead to the discovery of major changes in GHCN rural data?
Any urban sites that showed significantly different trends to the nearby rural sites were discounted in the temperature trend- you dont think that the overwhelming global body of scientists are stupid do you?
Very interesting. Compliments to all concerned for your tenacity, especially Peter and Dad for focusing attention on this issue.
Perhaps Anna V’s comment, above, is pertinent.
… Assuming that “value-added” data is still available. The dumping or destruction of “raw-raw” data (especially if it were shown to have been cooler) seems particularly incendiary.
Correction to above… “Assuming that something other than“value-added” data is still available.”
I just completed extraction and analysis based on the data and processing steps listed by GJG and JohnV. My data was ‘recent’ as noted by Anthony above. I extracted the data for years 1900 -2006 from GISS web site for the GJG list of paired locations, and removed the rows with 999 values in either Rural or Urban values as discussed above.
My results confirm JohnV’s numerical result, with one notable difference.
Rural temperature slope, DegC per year = 0.0057
Urban temperature slope, DegC per year =0.0107
Which gives I think
Rural 0.57/century (not per decade as listed above)
Urban 1.07/century
More results comparing Rural and Urban temperatures at the 28 paired sites, all in DegC:
……………….Rural ……..Urban
Minimum …….5.8………….7.2
Maximum ….22.4…………23.3
Average……..13.1…………14.3
As others have pointed out, the slopes of temperature change per time are probably more important than the averages, since there may be true siting differences in some pairs. Additional poking at the data showed no particular trends in the slopes based on warmer or colder areas in the paired sets.
*There are 6 of 28 (~20%) Rural negative slopes (cooling).
*None of the Urban temperature slopes are negative/cooling.
*Five of the Rural slopes are higher than the corresponding Urban slope (almost 20% of pairs), which also means that those five Urban slopes are lower temperature change than the corresponding Rural slope.
for what it’s worth.
Dave (Professional Engineer, PE – retired)
The retired part lets me poke around with data and spreadsheets from time to time. And of course, I am a daily viewer here at WUWT.
In my city, the Fire Dept operates a wundergournd.com weather station. There is a 2nd wunderground.com weather station located in a rural area, 4 miles outside of the city. The temperature reported by the Fire Station in real time consistently runs 3 to 5 degrees F above that of the rural station just 4 miles away.
Dave in Delaware:
Thanks for taking the time to do an independent confirmation and for posting the extra details as well. I used data from 1898 to 2009. That might explain the difference in the 3rd significant digit of the urban trends.
You mentioned “one notable difference” from my results. Is that the 1.07 vs 1.06 on the urban trends, or was there something else?
—
Bill Illis:
Do you believe me yet?
re JohnV (14:18:52) : You mentioned “one notable difference” from my results. Is that the 1.07 vs 1.06 on the urban trends, or was there something else?
Our numerical results are in good agreement for the updated process. A value of 1.06 vs 1.07 is right on the mark from my perspective.
In two of your earlier posts on this thread you listed different time scale units for your results
your initial post lists Per Century
JohnV (20:37:08) :
Rural trend: 0.81 degC/century
Urban trend: 0.94 degC/century
Difference: 0.13 degC/century
then later, after you had reworked your spreadsheet numbers, the new values are listed as Per Decade:
JohnV (11:02:42) :
After updating my analysis, I now get these trends:
Rural: 0.57 degC/decade
Urban: 1.06 degC/decade
The difference in Units (per decade) would also give perspective on Anthony’s comment at
Anthony Watts (10:21:03) :
BTW that 0.49C difference in trends you found is quite significant in the context of the generally agreed upon 0.74C that is claimed for the last century.
Since you listed ‘per century’ the first time, the ‘per decade’ may have been unintended in your later post. I believe that the units for the listed results should be per century in JohnV (11:02:42) : which was the ‘notable difference’.
regards,
Dave
Dave in Delaware:
That was my mistake. The second post should have been in degrees per century like the first.
Anthony’s comment about the significance of 0.49C is still true. A difference of 0.49 degC/century is very significant compared to 0.74 degC/century.
Of course, I don’t think this little sample of stations is the definitive word on UHI. But the method is reasonable so the results are worth something.
JohnV (14:18:52) :
Bill Illis:
Do you believe me yet?
Okay, I apologize.
I really just want to the real data to come out and sometimes it seems as though as some are just trying to further obscure the issues.
But in this case I was wrong.
I still would like to see an explanation of why the average of these stations (which seems to be a rather random sample) is higher than the average of the US in general.
i know it’s just weather, but here in north central Iowa, there doesn’t appear to be any warming at all since 1991.
using the average monthly temps reported on my utility bills, i plotted 2 graphs—– warm, ( june july august ) and cold ( december january february.)
the warm graph has no particular trend, including 1998 . MAYBE a slight neg trend from 2004 to 2009.
the cold graph, however, has a peak in 1998, returns to prior levels and remains fairly flat till 2004, when an easily seen decline begins and continues till present.
There are other threads on wattsup that say that the original raw data has been removed and what is called raw is massaged data.
I am really confused where this analysis falls, is it really raw raw data? All the acronyms are confusing too.
Bill Illis:
Thanks. I appreciate that. Apology accepted.
I am very careful about being honest and open so this was an important one for me. I will try to be less aggressive to avoid picking any more fights.
I don’t know why these stations give the trends they do, or why the analysis in the video shows a flat rural trend. I’m moving on though.
@ur momisugly DennisA
You know, that Munich Re is one of the funder of the PIK, where Schellnhuber and Rahmstorf come from, two of the greatest alarmist around ?
Anthony,
What happened to your UHI project and test run in Indianapolis?
I am new to the site and can’t find a followup. Transects look to me like a great idea. I am bothered by trying to use data from weather stations because of siting problems, especially since many metropolitan area stations appear to be at large airports.
FWIW, George Taylor, formerly a climatologist at Oregon State University said that an analysis of rural station data actually showed a statistically insignificant decline in temps while urban temps showed an increase. This was at a lecture about 10 years ago.
Since Taylor challenged the “science” behind global warming and wasn’t following the party line, the governor changed his job title.
What was the result of this?
Did anyone nail down the difference between the results Peter & his dad came up with, and those of JohnV?
Was it simply a question of the data set used? If so, that would of course be highly significant.
It was an interesting thread, with unresolved questions, which has gone completely quiet, and I’m not sure why.