Never before published paper on UHI and siting – Goodridge 1987

Philip Bradley says:
August 16, 2012 at 5:04 pm

More than changes in albedo, the changes in relativee humidity in comparisons between rural agricultural and rural non-agricultural I would expect to be substantially more influential. The forested areas can be quite dry in comparison to farmland under irrigation and direct insolation. Water vapor being a strong influence, the changes in time periods with irrigation versus without irrigation should show up quite strongly in the temperature and humidity observations. The urban aeas should also exhibit some of these changes as well with the inceases and descreases in the irrigatoin of urban landscaping and golf courses. A subset study of stations located in proximity to the golf courses past and present could be interesting.

faustusnotes: “My understanding of AGW science (admittedly limited) is that SST is predicted to warm more slowly than land temperatures.”
I’ll give you points for your stamina at dissembling in the face of adversity. I’ll simply reiterate the long and short of what’s in Goodrich and what you’ve finally been drug to kicking and screaming.
Based on a wholly arbitrary partition of the sets — without statement as to the how, why, what of that partition — we have one set correlate with SST and the other with population growth / total county population. And this, you finally, have the pained honesty to admit.
There are only two statements that are valid or invalid at this time. And only two for you to refute properly or be drug into sanity against your will:
(1) If this is prone to the Texas Sharpshooter fallacy — and that’s a legitimate concern — then so to the rest of the notion of ‘Global Average Temperature and Trends’. Specifically and especially if you start putting your thumb on the scale to ‘adjust’ and ‘correct’ for possible biome, human, instrumentation differences, or other nonsense.
(2) This warrants further research. The paper itself, that you’re screaming about, did nothing but sift existing data and look for various correlations. They were found, and without the help of a massive mathematical religion spewing tablets from a supercomputer. If this is seen as disconfirming or disconcerting to one religion or another? Suck it up. Duke it out like men that fancy pocket protectors and replicate or fail to replicate in differing data sets.

Nice to see a few more people stating that it is necessary to look at the CHANGES that have occured at a site. The 1970’s to 2000’s was generally a period of rapid urbanisation. Rapidly expanding urban settings even in so called rural areas. ANY site that was swallowed up by this urban expansion MUST experience a UHI effect, and without know the complete history of developement even within a couple of kilometers of the site, there is no way that you can account for possible urban warming effects.
And when I see BEST and others saying that there is no apparent UHI effect, then all I can assume is that they haven’t been competent enough to figure out how to find it, or have deliberately avoided finding it, to further their agenda and/or to keep the funding coming.

Pat Frank says:
JJ, I suspect we’re approaching a rapprochement.
There is a convergence.
We agree on your first point.
There it is. 🙂
Your second, “not so much” we almost agree, except that I’m willing to concede JG’s professional judgment.
Most people around here say “In God we Trust, all others show data.” Being Agnostic, I’m a notch tougher than that. 🙂
For your “not at all,” his Table 1 lists all 74 stations rural and urban stations. Table 2 again lists the stations, with the urban ones bolded, and gives location (lat., long., elev.) and regression slope.
None of which speak to urban/rural or the associated siting quality issues.
Three maps are given allowing correlation of site with population (Table 3).
They may allow correlation of site with population, but that wasn’t done. Sure, if a reader wanted to, they could take the station location data, and investigate each for surrounding population and other siting quality issues. Then they could run some stats and determine if the sites assumed to be “rural” based on their temp trend were in fact rural. They could do the same thing for the sites assumed to be “urban”, and they could publish those results, completing Goodridge’s paper for him. He could have done that, too. But he didn’t, and that is my point.
His two final figures demonstrate poorly sited stations.
Which is interesting in its own right, but those and the other examples given in the text are provided to demonstrate that those sites don’t fit the “urban/rural” classification assigned to them by the arbitrary trend break. What other sites also don’t fit? We don’t know.
Let me go a little further: Roger Pielke Sr. published a paper in 2002, … How is JG’s paper of lower quality than that peer-reviewed paper?
Leaving aside for the moment the logical fallacy of exonneration by association … a few things:
1) As you point out: “Table 2 was site description – lat., long., elev. years of data, and a short qualitative description of the area; like JG’s Table 2, though JG gave no description.”
How is JG’s paper of lower quality, with respect to urban/rural assessment of each site? Well, based on what you said, Pielke gave a site assessment for each site in his study, and JG didn’t . And what you (for some inexplicable reason) didn’t say is more damning still. Pielke’s Table II includes the current population for each site. Oh, and Pielke’s Table I gives the population history for each site, for the entire 20th century. Shame on you Pat.
2) Pielke’s paper is not about “urban/rural” differences in trend. It is about regionality and to a lessor extent seasonality. Pielke’s paper does not compare “urban/rural” groups of stations – or any grouping of stations, for that matter. His paper’s thesis is that individual sites are not representative of regional trends, so he compares all of the sites in a region against each other. Population isn’t even mentioned in Pielke’s abstract, and only appears in the text as an ancillary speculation on a couple of odd observations. It has no bearing on his results.
Soooooooooo … Pielke et al did a better job of documenting a factor (“urban/rural”) that has no impact on his results than did JG, who didn’t provide any documentation for the factor that was the basis of his thesis.
3) Pielke’s reasoning is not tautological, JG’s is. Comparing individual sites to each other and finding significant differences, and finding significant differences between individual sites and the average of all the sites in a region, is interesting. Finding that the trend of sites chosen because they have low trends is lower than the trend of sites chosen because they have higher trends is …
Recall that Philip Mullholland pointed out that JG’s manuscript had been cited by Tom Karl, et al., in their 1986 paper on “Urban Warming.”
So? Appealing to authority is a logical fallacy. Appealing to Tom Karl’s judgement is, well … not something that I would make a habit of.
So, JG’s manuscript is an extended abstract of the talk he gave at a professional symposium. We don’t know what he might have mentioned in his talk concerning the content of his judgments of station quality. However, one can’t expect an extended abstract to contain the sort of details one expects in a full paper.
If JG were truly taking the tack that you are relying on – that dividing stations on trend and looking for expected differences in “ruralness” was the hypothesis being tested, then the quantification of those differences in “ruralness” would have been the results of the paper. One does expect a minimalist abstract to present the results, let alone an extended abstract. Kind of goes to the definition of an abstract.
One also doesn’t expect hand-waving arguments to fly at a Society-sponsored scientific meeting.
Well, I have seen all sorts of nonsense flying at scientific conferences, expected or not. Too, we don’t know if what was presented at the Society sponsored meeting did fly. It may have been panned. Either way, it says nothing about what was presented. The only thing we have before us that descrbes the content of the talk is this paper, and this paper is lacking.
You are denying the independence of SSTs.
No, I am denying the independence of the method of dividing stations by trend from the “finding” that the trends of the stations so divided differ according to the method of their division.
There is no reason to think that SSTs would correlate with any arbitrarily divided set of temperature trends.
Sure there is. Don’t be silly. Given the heat capacity of the ocean, SST trends are expected to be more or less flat, regardless of your position on ‘global warming’. Pick a group of *anything* based on the flatness of its trend, and you would expect it to correlate with SSTs. And you won’t be wrong.
There is plenty of reason to think that SSTs will correlate with unbiased local land-surface temperatures.
You are assuming “unbiased”, but you aren’t grouping based on documented bias. You are grouping on trend and calling the group based on the trend that you would expect of an unbiased group “unbiased”, and calling the group based on the trend that you would expect of a biased group “biased”. Upon examining what you have done, you then “find” that the high trend group has a higher trend than the lower trend group, and you pretend that this says something about bias. Sorry. No. You’re just regurgitating your assumptions.
The correlation of SSTs with one trend, but not the other, validates the rural/urban distinction,
No. That line of reasoning is a form of confirmation bias.
“… in that the SST correlation implies stations that are free from urban heating.
No, the SST correlation confirms stations that are free from *any* significant net heating, “urban” or otherwise. Given that the stations in question were chosen because they didn’t demonstrate net heating, this is not surprising. It is tautological.
Such stations were found to be JG’s rural-designated stations. That finding was not imposed by his prior choices.
Uh, yes it was.
The same conclusion applies to the similar rural trends found by others.
I concur. Comparison to other “rural” trends is similarly flawed. Essentially, you are saying that the “findings” of JG’s paper are not in JG’s paper, but that his findings are to be found in other people’s papers. Silly.
As I stated above to another poster, there are a couple of different ways to legitimately skin this cat:
1) Divide stations based on some measure of “ruralness”, and compare the trends between those groups for evidence of difference. In this case, differences in trend are the results.
2) Divide the stations based on a measure of their trend, and compare the “ruralness” between those groups for evidence of difference. In this case, differences in ruralness are the results.
Either of those works, but JG didn’t do either of those. And it isn’t an ala carte deal. Using the methods from 2) with the results from 1) – which is what JG actually did – is simply not valid.
These sort of circular reasoning fallacies can sometimes be difficult to detect. That really isn’t the case here. It is pretty obvious.
JG said that he divided his groups based on the magnitude of their temp trend. He says he called the low trend sites “rural” and the high trend sites “urban”. Listen to him when he says that. If you want to clearly see what he did next, just go thru his paper and everywhere you see a term like “rural site” or “rural temperature” in reference to his data, replace it with “low trend site” or “low trend temperature” and every time you see a term like “urban site” or “urban temperature” so used, replace it with “high trend site” or “high trend temperature”. So stated, the ‘results’ aren’t so compelling.
The known-to-be rural trends (i.e., Diaz) have had their siting verified. Their correspondence with JGs trends implies the correctness of his distinction.
What correspondence? JG doesn’t demonstrate any correspondence with Diaz. What known-to-be-rural trends in Diaz? JG doesn’t even make that claim. It wouldn’t matter if he had, for the reasons shown above, but he didn’t even do that. You are seeing things that aren’t there.
Let’s note that JG wrote that he had inspected the sites. He just did not give any specific details.
Lets note it correctly. JG said that he had “viewed most of the sites”, and he said that the classification of that subset of the sites was “fairly close to reality” based on his recollections of having viewed that subset. Nothing objective or quantified. “Fairly close” is not only subjective, but a very weak commitment at that. Lipstick does not a prom date make. It simply annoys the pig.
He did provides maps, locations, and populations, however.
Well, maps and locations is the same parameter, and he did not provide site populations. If he had, people might reasonably have expected him to have used those populations as his “urban/rural” criterion.
He did provide county populations, but he did not use them to determine “rural/urban” for his sites. To the contrary, his list has many instances of both urban and rural designated sites within the same county, rendering county population irrelevant to the question of “urban/rural” insofar as JG’s paper is concerned.
As it stands this paper is seriously flawed. It isn’t necessarily beyond fixing, tho. Goodridge didn’t do it right, because he didn’t have the resources to get the data needed to do it right. Anthony probably has those data, having had access to a resource that Goodridge did not: free labor sourced from the internet. It is likely that the sites referenced in JG’s paper have been visited and classified by the Surface Stations project. It would be interesting for someone to start over with JG’s “implied hypothesis”, split the sites on trend, and use Anthony’s data on siting quality to determine if there are differences … it wouldn’t be the first paper that took 25 years to finish.

jorgekafkazar says:
August 16, 2012 at 5:56 pm
…
[I’ve lost this comment twice, already, when I clicked on a chart while trying to verify something,

Lazarus Add-On. Lose nothing. Ever. Save yerself scadoodles of time an’ aggervation. Promise.