Talk:Instrumental temperature record

From Wikipedia, the free encyclopedia

WikiProject Environment
Portal		 This environment-related article is part of the Environment WikiProject to improve Wikipedia's coverage of the environment.
The aim is to write neutral and well-referenced articles on environment-related topics, as well as to ensure that environment articles are properly categorized.
See WikiProject Environment and Wikipedia:Contributing FAQ.
B This article has been rated as B-Class on the assessment scale.
Archive
Archives
/Archive_1 /Archive_2 /Archive_3



Contents

[edit] Old general discussion

I've (William M. Connolley 20:03 Feb 12, 2003 (UTC)) moved 2 paras to the end. These are:

the intro that someone copied in, viz:

"The historical temperature record shows the fluctuations of the temperature of the atmosphere and the oceans throughout history. Climate scientists generally agree that Earth has undergone several cycles of global warming and global cooling in the last 20,000 years, with the average air temparature fluctuating within a range of about 3 Celsius degrees (5 Fahrenheit degrees), over this time period."

This is mangled info. Someone can straighten it out if they like. If you look over the last 20 kyr, the biggest signal you see is the end of the last ice age - so the stuff about little cycles is then in the noise.

There are various sub-cycles/sub-signals, of ??1500 year ish?? periodicity; and their are the D-O events etc etc. But the above para mangles that. *Also* it fits rather poorly with the emphasis of this subsection-now-a-page, ie on the last 150 or 1000 years - so it shouldn't be up there in the intro.

I've also pushed

"In January 2002, scientists released data showing that Antarctica had grown about 25% (???). Some editorial writers claimed that this contradicts the expectation that rising temperatures should cause the ice cap to shrink. However, the scientists studying the situation in the Antarctic who released this data point out that local cooling in some areas is consistent with an overall trend of global warming and say that "the ice-sheet growth that we have documented in our study area has absolutely nothing to do with any recent climate trends."[9]"

into the misc section. The first sentence is junk. If its to stay, someone has to find a decent ref to what its supposed to mean. Mind you, ref [9] is nice and its a pity I've misc'd it too...

The IPCC says that it has corrected the land station data to account for the urban heat island effect. To do: find and summarize their correction technique.

The comment above has been around for about a year, and still no one has shown me where in any IPCC report they have explained how they "account for" urban heat islands. So I'm inclined to say rather:

Critics of the IPCC report note that it fails to explain how it accounts for the urban heat islands. These critics argue that the heat island effect correlates with land-based thermometer readings better than the global warming theory espoused by the IPCC.

... or something along those lines. Work with me here, folks. Let's make an informative and neutral article. --Uncle Ed

That looks fairly reasonable - I'd modify slightly:

The IPCC report does not explain how it accounts for the urban heat island effect - increased warming due to proximity to major cities. The heat island effect, if not properly accounted for, would tend to increase the amount of apparant warming.

Martin

(William M. Connolley 09:46 Feb 13, 2003 (UTC)) There is at least a problem in the wording here. IPCC doesn't do research, it reports other peoples. But apart from that... see section 2.2.2.1 [1]. In particular:

These results confirm the conclusions of Jones et al. (1990) and Easterling et al. (1997) that urban effects on 20th century globally and hemispherically averaged land air temperature time-series do not exceed about 0.05°C over the period 1900 to 1990 (assumed here to represent one standard error in the assessed non-urban trends). However, greater urbanisation influences in future cannot be discounted. Note that changes in borehole temperatures (Section 2.3.2), the recession of the glaciers (Section 2.2.5.4), and changes in marine temperature (Section 2.2.2.2), which are not subject to urbanisation, agree well with the instrumental estimates of surface warming over the last century. Reviews of the homogeneity and construction of current surface air temperature databases appear in Peterson et al. (1998b) and Jones et al. (1999a). The latter shows that global temperature anomalies can be converted into absolute temperature values with only a small extra uncertainty.

Errr... shouldn't all this go into the UHI page?

Yep. And then summarised here. :) It all seems rather a lot of work... :-/ Martin
William, the large quote above (0.05°C over the period 1900 to 1990) seems at first reading to answer Martin's other question: are heat islands causing global warming. My question is different: are temperature readings taken within heat islands giving a false impression of global warming. That is, (1) if a city gets 0.8°C warmer, and this warming is averaged in with all other temperature differences, I think this would be a statistical error. What do you think? Also, (2) if cities get much warmer, suburbs get kind of warmer, rural areas get a bit warmer, and uninhabited areas don't get warmer at all, what would this tell us? (Not saying that's the case for now, just asking what this would tell us if it were so.) --Uncle Ed 17:19 Feb 13, 2003 (UTC)
AFAICT, "globally and hemispherically averaged land air temperature time-series" - I think this phrase is referring to the temperature readings rather than the actual temperatures - IE, it answers your question... Oh, I'm copying some of this stuff to the urban heat island page. Martin
There is an answer to UE from IPCC. Essentially, you can (if you wish) separate out the obviously-likely-to-be-affected stations if you like, but it makes little difference. Reasons include: cities are small areas anyway; the trends from cities (etc) don't in fact differ substantially from the trends without them; in fact the trends over city areas agree quite well with the dreaded MSU... I'll try to find this and add it in, since its clearly a concern (William M. Connolley 21:31 Feb 13, 2003 (UTC)).

I have started studying a paper on the temperature record in the USSR. The writers find no warming trend in rural stations and hint (or imply) that other researchers have selectively chosen data to fit their "warming" views. [Read it yourself] and decide. --Uncle Ed 17:52 Feb 13, 2003 (UTC)

Its worth reading (William M. Connolley 21:36 Feb 13, 2003 (UTC)). I've had a brief look before. Of the 2 stations I picked to check his analysis, one had the jump he claimed (according to objective statistics) and one didn't, or it was impossible to tell: the trends he claimed as implausible against "neighbouring" stations were from places 100's of km away, and in different exposures: near the sea or not.

Martin and William,

I'm not sure either of you is getting my point. I am not wondering whether a few hot cities are making the whole world hot.

I am wondering whether a large number of the temperature readings from weather stations in and near rapidly warming cities, when averaged with a relatively small number of temperature readings from rural and remote stations, are giving a false impression of global warming. That is, it might be that (A) the only parts of the world that are warming are the urban heat islands and (B) the only reason these are heating up is because cities absorb and generate heat; rather than (C) that carbon dioxide, etc. is causing worldwide warming.

Do you understand my point? (I'm not asking whether you agree with my point of view, but only whether my English is clear.) --Uncle Ed 22:33 Feb 13, 2003 (UTC)

I believe so - let me paraphrase to try and prove it.
* Because of the urban heat island effect, cities are warming up more than the surrounding countryside
* Thermometers are recording an increase in temperature
* Many thermometers are located near urban heat islands
* Therefore, the temperature increase recorded by thermometers may overestimate the actual temperature increase of the climate as a whole.
To draw a parallel, one shouldn't put the thermostat in one's house next to a log fire, because in that case the thermostat will overestimate the general temperature of one's house.
My reading of the IPCC report is that the distortion introduced by the urban heat island effect is, at most, 0.05°. In other words, if we had located our thermometers away from urban heat islands, they would have recorded 0.05° less temperature increase over the period 1900 to 1990. Of course, this depends on whether you trust the research cited by the IPCC... Martin
I mostly agree with Martin. Have you read the bit about marine and borehole temperatures? This does a lot to counter your point. I also think you're wrong to suppose that, numerically, urban reading predominate. Martin: note that strictly speaking IPCC reprots that UHI leads to at most 0.05 *uncertainty*. They don't (I think) explicitly state that this is necessarily in the warming direction.

Thank you, both, for helping me to feel understood. Now I'll have a G-R-E-A-T weekend! ^_^ (Uncle Ed)

[edit] Preliminary results

The section:

Preliminary results as of September 16, 2007 indicate that the best rated and worst rated stations have similar overall temperature trends. Steven McIntyre stated that "between the 1930s and 2000s, the differences are material, but for comparisons over the past 30 years in the U.S., the differences are less."[1] He also stated that "CRN1 [class 1] stations turn out to be skewed to the east, especially the southeast, while CRN5 [class 5] stations are skewed to the west." He argued that, when they are compared to the highly rated station results, "The profound differences between NOAA and NASA [overall] results obviously point to substantial differences in their adjustment methods" and "the NASA [temperature] history for the U.S. looks more reasonable."[2] Co-bloggers at McIntyre's website have argued "There is good agreement between" the GISS, CRN1 plus CRN2, and CRN5 station records.[3]

has been removed. I agree that it shouldn't be in the article (or, at least, should not be in the article right now). Setting WP:SPS aside just for the sake of arguement, the results are "preliminary" and have not been subject to any kind of peer review. Revolutionaryluddite 21:54, 16 September 2007 (UTC)

This is inconsistent - none of the Watts stuff has been PR'd William M. Connolley 08:49, 17 September 2007 (UTC)
I agree that the PowerPoint section should also be removed. Revolutionaryluddite 20:07, 17 September 2007 (UTC)
Every single data point was offered to the respective station maintainer. They had the right to correct any errors and still have that right. That's effective peer review and built into the data gathering process. Since nobody else has done the data gathering, there's nobody else who could actually PR this stuff but the people actually on site, either the survey taker or the site maintainer. It's nonsense to say it's not been peer reviewed when the data gathering and checking process is this open and any site maintainer can correct errors easily and quickly.
WP:SPS is a red herring as the data I included (late October) was only put out for public consumption after the UCARS talk, the data was released for two weeks of review, and nobody came and offered any corrections. So we have two levels of review, certainly not academic journal level (unless you're on the Lancet fast track perhaps) but certainly better than the type of self-published source that is intended to be stopped by WP:SPS.
I didn't want to make a big deal out of this data and give it undue weight. I put in what I believed were appropriate caveats as to the incompleteness and preliminary nature of the results. But if they hold up (and they do seem to be broadly holding up) somewhere north of 50% of the USHCN network will end up being CRN category 4 or 5 stations with noise exceeding signal. I don't believe that *anybody* including most skeptics would have thought that things were that bad and the strong nature of the data so far and the post conference review led me to believe that fair minded editors would let this in. TMLutas 18:14, 26 October 2007 (UTC)
"Peer reviewing" a data point is not the same thing as peer reviewing an analysis of several data points. Ben Hocking (talk|contribs) 19:37, 26 October 2007 (UTC)
We're not talking rocket science analysis here. You apply the CRN standards to the survey results. They get you a CRN rating. My understanding is that the individual station maintainers can challenge the CRN rating as well as the photos, diagrams, and survey forms. That's the data point. The only thing not peer reviewed is the arithmetic, how many 1, 2, 3, 4, and 5 rated sites are there. It's a pretty trivial exercise since everything is available on site. Are you saying Watts doesn't know how to add or is adding dishonestly? What else isn't reviewed? Sure, you could say that it would be original research to actually tote up the ratings but come on. TMLutas 02:34, 28 October 2007 (UTC)
I doubt it is that simple. The fact that they can say the magnitude of the 'error' tells me that something more complicated is taking place. Brusegadi 02:43, 28 October 2007 (UTC)
More to the point, it has become obvious that TMLutas doesn't understand what "peer reviewed" means. Raymond Arritt 02:46, 28 October 2007 (UTC)
What I'm getting at is that peer review is not the be all and end all of independent editorial review. Especially in cases of simple observation, peer review is actually less useful than review using Watt's method. Peer review is a crutch, a simple standard that everybody can agree on. It is not the only standard to determine reliability and when this particular project is peer reviewed, the peer review is very likely not going to provide an awful lot of additional reliability. How could it when the reviewers have no way of knowing whether the data is correct or fabricated? Only the site maintainers know that and they have already reviewed the data.
In cases where there's a little data gathered and a lot of interpretation, peer review is the best method. I am suggesting that in cases where what's going on is mostly data gathering such as this study, independent review of all data points is a much higher level of meaningful review than peers who are not going to be going across the fruited planes to check out 1200 sites independently. When the only analysis in a project is the equivalent of =SUM(x) in Excel I cannot understand the blind insistence that only peer review matters. TMLutas 00:34, 29 October 2007 (UTC)
Brusegadi - No, it really is that simple. The CRN created an objective system for assigning error ratings, a 5 point scale with associated temp error magnitudes. The study simply takes the CRN standard and applies it. If you're a category 5 station, you're doing something incredibly stupid like sticking a temperature sensor on a black tar roof right next to an AC exhaust vent and thus you get assigned the CRN error factor for a category 5 site (5C or greater). There's no actual error measurement involved in the study, just a measurement of distance to the nearest concrete, artificial heat source, or other factor identified in the CRN standard, documenting it, getting the site maintainer to sign off that the surveyor is accurately describing the site, and then submitting it to Watt's website. Now you can go attacking the CRN standard but that's a pretty steep hill to climb and very different than taking on Watts' project. I can understand the confusion but you can certainly double check that this is what the CRN system does independent of my explanation and judge from there. TMLutas 00:34, 29 October 2007 (UTC)
Then, where does the error above 5 Celsius come from? Being able to say that X station has an error >5 seems pretty non-trivial to me, at least in the context you speak of. How do they determine that? Brusegadi 01:37, 29 October 2007 (UTC)
CRN calculated that if you do certain forms of site stupidity, you know that your temps are going to be off by more than x, x changing depending on CRN category and the observed site stupidity determining the CRN category. It's a simple value pair. If you are a category 5 site, the error rating is =>5C. If you're a category 4 site, the error rating is =>4C, and so on. Watts is compiling surveys, diagrams, and pictures and applying the CRN rules to determine which category applies. The temperature error rating sort of comes along for the ride. That's very mechanical but nobody seems to be challenging the CRN rules themselves. TMLutas 13:53, 29 October 2007 (UTC)
Last time I looked at the site there was a heavy bias towards urban stations. (Not surprising considering the fact that this is a mostly volunteer project, so stations near high population-density areas can expect to have more volunteers.) That's one thing that the peer-reviewed process is meant to catch. Ben Hocking (talk|contribs) 14:00, 29 October 2007 (UTC)
Since the project is going to go through all 1200+ sites, urban bias is a passing problem. The effect is real though amd worth discussing. The bad stations have dropped in % from the mid 70s to the high 60s to my observation. That's why I'm interested in working out how low the 4+5 total % has to be before it's no longer considered significant. 68% at 1/3 done equals at least 22% bad even if zero stations that are 4/5 rated are found for the rest of the project. Is 22% of the network being 4 or 5 on the CRN scale a big deal or is it a small thing? If we agree that 22% USHCN bad sites is a significant result, the project has already met the threshold of whether it should be included. We still need to wrestle with standards of independent review (peer review v. USHCN site maintainer oversight) but we'll have settled a big issue, when does finding bad weather stations stop being an isolated, unimportant instance and when does it start being a serious challenge to conventional studies that depend on these weather stations for some or all of their conclusions. Then we'd have an easily followed consensus on when the preliminary stuff should start getting a mention. So can we start with a number? Let me toss out 10% 4/5 rating as already being a problem. If you disagree, give a better reason and a better number. TMLutas 03:17, 30 October 2007 (UTC)
Let me toss out zero peer-reviewed publications as being a problem. (Yes, this is becoming tiresome.) Raymond Arritt 03:32, 30 October 2007 (UTC)
Yes, you've said this. The question is whether you'll say anything else. The standard for inclusion is *not* peer review. It's independent editorial oversight. The data being discussed for inclusion does not have peer review but has an alternative system of independent editorial oversight. I have explained the system. You have not progressed beyond your original point other than to call my attempts at working towards a consensus 'tendentious'. I sent you a note on your page to sort that accusation out too. You have yet to respond constructively there either. Either engage fully to work things out or stop the broken record routine. TMLutas 13:59, 30 October 2007 (UTC)
Sounds like a problem to me. As RA points out, however, without peer-reviewed publications who are we to say how big a problem it is or whether it even is a problem. Do you see now how this is not just a matter of whether or not the individual data points are accurate and why peer-review is relevant? Ben Hocking (talk|contribs) 13:48, 30 October 2007 (UTC)
We are supposed to keep out stuff that doesn't have independent editorial oversight or at least give such material much higher scrutiny. Peer review is a handy way to demonstrate independent editorial oversight; I agree that it's quite useful. This does not establish that it is the sole method of oversight. I would say that in a certain number of cases, probably small but nonzero, alternate arrangements are sufficient for inclusion in wikipedia.
1. The material should be important, having large implications in the field. - probably check
2. The material should be transparent, easily checked by anyone (open source "many eyes" method) - check
3. The material should have independent checkers not associated with the project, ideally people who have an interest contrary to the way bias is likely to run - check
4. The material should have a good prospect of transitioning to more conventional editorial insight, ie this is a patch that is worth applying because of the importance of getting an early heads up disseminated widely. - maybe check
5. The material should be unique, ie there shouldn't be peer reviewed alternates available. - check
Now I think items 1 and 4 are satisfied but I freely admit that they're the weakest of the list. One can say that the list is insufficient. Good, give me some other conditions to add to it. But you can't say that the lack of peer review sinks it because that's not part of the wikipedia rules, policies, or guidelines. Peer reviewed stuff is to be preferred. But nobody else has done the physical review of the USHCN (and no doubt later, other networks). The unique nature, importance, transparency, and likelihood of future publication make the case for overriding the usual procedure. This is not an easy set of conditions to meet. If there were serious unanswered methodological questions it would sink on condition 2 (and condition 2 is the direct answer to your immediate question of how are we to judge). So far, there don't seem to be any.
I sometimes don't like Wikipedia's actual rules but I recognize that they need to be defended in order to preserve the viability of the project. Peer review only could be one of those rules but it currently isn't. If editors want it to be, they should change the rule articles. But they don't do that. They just revert. And when there are enough reverters or deleters violating actual wikipedia standards it *can* be made to stick for a time. It isn't good for wikipedia though and sets up a backlash. The backlash is coming. There's an active boycott of wikipedia financing (not over science articles though) and that sort of thing either has to get addressed or it's going to grow.
I want to be clear. I'm including this edit and fighting for it in talk based on its merits. This episode does, however, fit into this larger pattern of people not bothering with going through the process of changing the rules but simply using energy and numbers to impose by fiat alternative rules so I would prefer that both the small and larger issues be addressed. If people don't want to open the larger can of worms I can understand that and we can just talk about this particular edit so long as we stick to the actual rules. Wikipedia has no common law. TMLutas 14:31, 30 October 2007 (UTC)
TML, please read Wikipedia's essay on tendentious editing: "On Wikipedia, the term also carries the connotation of repetitive attempts to insert or delete content which is resisted by multiple other editors." You're pressing the same point over and over again, not just on article talk pages of articles but on the user talk pages of individuals who disagree with you.[2][3][4] Raymond Arritt 15:07, 30 October 2007 (UTC)
Read it, doesn't apply, mostly because I'm currently not trying to actually edit anything. One cannot be tendentious over an edit one has conceded isn't going to make it at the present moment. I'm trying to figure out how to form a consensus on when preliminary results flip over into stuff that qualifies under WP:RS and apply it to the edit so the next time I put the thing in, the edit will stick. Contrary to what you seem to think, I don't actually like edit wars. On a fairly fast moving target like the Watts project I believe the best bet is to form a consensus on when to include it in accord with WP:RS. TMLutas 02:57, 31 October 2007 (UTC)
The consensus is to consider including it when it's been published in a peer-reviewed journal. You reject that consensus, so you're trying to wear the rest of us down by sheer repetition and weight of verbiage. Raymond Arritt 03:20, 31 October 2007 (UTC)
You cannot, by consensus, change WP:RS without actually changing WP:RS. The reason is that it's a false consensus. The actual rule being applied is not what is being stated, that scientific articles can only have peer reviewed sources in them. If it were, there would be a gutting of Wikipedia's science articles, especially stuff on the cutting edge. Go change WP:RS, or at least try to and I won't be using this argument anymore. We've bumped heads enough in the past that you know I don't persistently go against actual rules, policies, or guidelines. My prediction is that if someone were to actually try to change WP:RS along these lines, they would get slapped down, hard and thus this false claim of consensus is raised but the rule is never actually changed because the guys monitoring *that* article would not appreciate or agree with the change in rules that RA and others propose. TMLutas 21:40, 4 November 2007 (UTC)
I agree with Raymond William M. Connolley 22:06, 4 November 2007 (UTC)
Fair enough WMC, what, exactly do you agree with? TMLutas 15:37, 7 November 2007 (UTC)

There are lots of problems here. TML seems to be pushing "the stuff is effectively peer reviewed": "Every single data point was offered to the respective station maintainer. They had the right to correct any errors and still have that right." - who says so; and is there any evidence that this has actually been done? Edit comment of "421 results, 421 ind reviewers" appears to suggest that TML believes ?all? 421 points have been reviewed?

Also the T errors: being assigned (even by an experienced observer, let alone an unknown) to a cat5 does *not* mean the T is in error by 5 oC. How could it? The error must vary by time of day, etc etc. At best it can mean that an error of this order is possible. Thus this edit [5] is unacceptable.

And "The unique nature, importance, transparency, and likelihood of future publication make the case for overriding the usual procedure" is completely wrong William M. Connolley 14:40, 30 October 2007 (UTC)

Let me start by conceding a small error. I should not have said "That's effective peer review and built into the data gathering process" but rather "That's effective independent editorial oversight and built into the data gathering process". Mea culpa, mea culpa, mea maxima culpa. Of course peer review is not the standard by which one includes information according to WP:RS. Sources with effective independent editorial oversight qualify just fine. The exact words are "Reliable publications are those with an established structure for fact-checking and editorial oversight". That exists for Watts' efforts.
Further relevant text in WP:RS with which I have no problems

Academic and peer-reviewed publications are highly valued and usually the most reliable publications in areas where they are available, such as history, medicine and science. Material from reliable non-academic publications may also be used in these areas, particularly if they are respected mainstream publications. The appropriateness of any source always depends on the context. Where there is disagreement between sources, their views should be clearly attributed in the text.

In other words, the rote reliance on including only peer reviewed articles is simply does not conform with WP:RS as written and I would suggest either dropping that de-facto standard or get busy rewriting WP:RS. If there was some sort of published peer reviewed article regarding a physical review of surface weather stations it should absolutely be preferred over the preliminary stuff at surfacestations.org. Do such papers exist? I don't think they do. So absent publishing the most reliable data available on the subject should be included at least in some form. Otherwise the article leaves a false impression about the current state of things.
You are correct that sticking a weather station in the exhaust vent stream of a 2 ton commercial AC unit does not mean that it will be off by exactly 5C all the time. I never said it did. Watts never said it did. The CRN never said it did. Straw man much? What is being said is that if one were to do such a daft thing, you're going to be at least 5C off and quite possibly more. The US taxpayer funded CRN to come up with such a standard and they have. That you don't like it is neither my fault, Watts' fault, nor should it be an impediment to reporting application of the CRN standard in a relevant wikipedia article. You can certainly provide the proper nuance and improve my edit. By all means go ahead. But that wasn't what happened. All of my edits on the subject were reverted, not improved. That's not supposed to be a first choice but it seems to be the first choice in this sub-community.
That each data point (defined as filled survey form, diagrams, and photos) is reviewed by the site maintainer is part of the design of the survey. If the volunteer hasn't done it, Watts says he's not publishing it. I've no doubt that if a site maintainer were to say that he wasn't consulted and Watts wouldn't let him correct the record it would be big news. This is a dog that has not barked. So yes, I think it's fairly safe to say that Watts is not committing academic fraud, or at least that's the point we should start from.
The last bit of your comment WMC is too general for me to reply to without an unproductive level of snark. All wrong? Really? I guess I should be satisfied you had as much intellectual engagement as you did prior to descending to a less productive style. I'll take half a loaf on that one and await your answers to my replies. TMLutas 02:57, 31 October 2007 (UTC)
Most of that is simply repetition of errors. There is no case here for overriding procedures. There is no evidence that oversight has actually occurred (the issue is not site maintainers asking for corrections that haven't occurred: the issue is whether any site maintainers have ever bothered to check Watts people). Reliable publications are those with an established structure for fact-checking and editorial oversight - indeed. There is no evidence for Watts efforts having any such. If you care to be more polite, I could try to explain the T errors stuff you've failed to understand in more detail William M. Connolley 22:05, 4 November 2007 (UTC)
You have to either assume that the project is lying (Watts' surveyors don't actually follow procedures and explain and clear things in advance with site maintainers) or that out of 421 surveys published so far the number of people who said 'yeah, let me see what you've got before you leave' is insignificantly small or nonexistent. That doesn't pass the laugh test with me especially as Watts' efforts are not exactly low profile. No evidence indeed, it's hard to find evidence when you assume so little good faith on the project's part.
You can explain your objections on the T errors stuff or not as you please. If you refuse to explain it, your objections on those grounds should not be taken into account. Fits of pique should not be catered to in a debate over scientific evidence and argument from authority remains a logical fallacy. "I know you're wrong but I won't demonstrate how" is not a sustainable method for maintaining wikipedia. Do you even consider how the other side in any debate could use the same tactics against you? TMLutas 15:37, 7 November 2007 (UTC)
"You have to either assume..." No one has to assume anything. Given that it is not a reliable source, the assumptions are made for us by the foundations of wikipedia. Brusegadi 17:11, 7 November 2007 (UTC)

[edit] Wikiproject Earth

Hello i have recently proposed the Wikiproject Earth. This Wikiproject`s scope includes this article. This wikiproject will overview the continents, oceans, atsmophere and global warming Please Voice your opinion by clicking anywhere on this comment except for my name. --IwilledituTalk :)Contributions —Preceding comment was added at 15:39, 30 March 2008 (UTC)

[edit] Preliminary results 2

The surface stations group stuff keeps piling in and the trend of many bad stations persists. Class 4+5 categories are now 69% of the 534 stations currently recorded. That's 43.73% of the entire set of stations. The question remains when the blinders will come off and those for whom these results are inconvenient to their own theories will admit that the data should be included in the article.

Here's a proposed edit: "A private open source effort called surfacestations.org is ongoing to rate the entire USHCN network according to NOAA's guidelines with follow on targets to rate the entire world. At present time, x% of the US network is complete and y% of the surveyed sample is rated as having an error rating higher than 20th century global warming observed by those instruments. Editorial oversight is maintained by the project rule providing that any individual site survey may be challenged and fixed." TMLutas (talk) 13:36, 2 May 2008 (UTC)

Encourage them to write something up for peer reviewed publication. Seriously. Dragons flight (talk) 14:53, 2 May 2008 (UTC)
You're missing the point. I think that the biggest ever physical review project of the instrument temperature record is, in itself, something that qualifies for mention in a wikipedia article on the instrument temperature record. I think that papers are likely (if they already haven't started) to come out of this but the project itself is noteworthy and I find the false requirement (see two sections up) of peer review that some keep bringing up to be... less than helpful for the goal of a complete, accurate, NPOV article on this subject. I *have* encouraged the project people to make sure that they have independent editorial review in my brief correspondence with them but was assured that I was already late to *that* party.
Peer review is a comfortable crutch, a nice, very functional example of independent editorial review that usually works quite well (though everybody can think of at least some instances where it's been worthless). It is not the sum total of independent editorial review nor is it adequate for dealing with something like this where you're only likely to get papers about the analysis some scientist did of the data coming out of the project, not the project itself. We can do better and the rules specifically allow us to. It would be foolish to put in artificial constraints above and beyond the rules without changing the actual rules (which nobody seems interested in doing). TMLutas (talk) 12:20, 3 May 2008 (UTC)
Correct. Peer review is not required. As was pointed out to me by User:William_M._Connolley on the talk page of Urban_heat_islands. In fact he made that very point to defend the use of a simple newspaper article in supporting the view that he holds which is -- UHI do not bias the temperature records. Therefore -- regarding Watts site -- I see no reason why the same logic cannot apply. Whether or not Watt's work is peer reviewed in a formal journal is not really the point. As an online encyclopedia, WP:NPOV is required but that does not mean peer review is required. SunSw0rd (talk) 15:31, 3 May 2008 (UTC)
I find that funny. If you look above, you may detect WMC taking the opposite point. Or am I reading his prior remarks on this page too closely? I haven't checked the UHI page and will take you at your word on your characterization. TMLutas (talk) 05:01, 7 May 2008 (UTC)
The simple fact is -- it is not valid to disqualify a point that is backed up by a secondary source, if the only objection to that secondary source is that it is not a peer-reviewed publication. My point is -- peer-review is "nice to have" but is not a requirement. Some examples of secondary sources discussing surfacestations.org may be found here:
Surface Stations and here:
Anthony Watts Has An Excellent Summary Of His Research So Far and here:
December 2007 Session ‘The “Divergence Problem’ In Northern Forests and a more critical link here:
The Surface Stations Project: Science Auditers or Enviro Vigilantes? and a semi-critical link here:
Pouring Salt On Climate Critics "Contaminated" Wounds and a news article here:
Meteorologist Documents Warming Bias in U.S. Temperature Stations
Hope these links are of use. SunSw0rd (talk) 17:42, 7 May 2008 (UTC)
Very interesting - but not a single of these are a reliable source, most are self published sources/blogs and one is a political advocacy site. The only one that could be argued as such (reliable source) is climatesci - but that puts too much weight on Pielke Sr.. The opinion of the blogosphere/political think-tanks is (while interesting) not usable in science. --Kim D. Petersen (talk) 18:15, 7 May 2008 (UTC)
To expand, Pielke Sr's opinion is already noted in the article, in what is (at the very least) close to undue weight. --Kim D. Petersen (talk) 18:17, 7 May 2008 (UTC)
Well one must be careful when indicating whether they are, or are not, reliable sources. For example, in this case, we are talking about references to a website run by an organization. Therefore, the typical references in this era are of course mainly going to be blog entries. But as long as the blog owners are not writing about themselves, and are writing in their field of expertise, like climatesci, are reasonable. As for heartland, the story was under the area of environmental and climate news. You may argue that they are a "think tank" or "political advocacy site" and their "opinion" is "not usable in science". Says who? In fact, you and I had a discussion on this very point on the UHI talk page -- and there and then YOU took the diametrically opposite position. The point was whether David B. Sandalow was a reliable source. I asserted that (a) he was quoting an anonymous source and (b) he is not a scientist but a lawyer and (c) that he posted at a political advocacy site. And you responded to me and I quote: "That would only be relevant if you have any reason to assume that the Sandalov reference is unreliable. The threshold for inclusion in wikipedia is attribution to reliable sources - not truth."
So. IF David B. Sandalow writing for an advocacy organization is acceptable, to you, for a citation in UHI, I don't see how you can stand by your current statement regarding heartland. If one is good, so is the other. Otherwise exclude them both. But otherwise it appears you accept the one if it backs a position you support, then reject the other when it argues against a position you support -- but both sources can be criticized equally and should either both be excluded or both accepted. SunSw0rd (talk) 17:44, 8 May 2008 (UTC)
Try to separate different articles and discussions - the context is rather different. Focus on thís article, this discussion and the policies.
I didn't say that climatesci in all cases is unreliable (it is here for what you want to use it for - but thats another issue) - but because it would give undue weight to Pielke Sr.'s view, which are already overrepresented on this page. I also suggest that you read through and understand WP:SPS. Weight and SPS are the two major problems you have here. --Kim D. Petersen (talk) 18:31, 8 May 2008 (UTC)
Actually, the problem is very much double standards, at least in some editor's views. Rules are either convenient excuses to bash something you don't like or objective standards that we're all to follow. I rather like the latter vision myself. TMLutas (talk) 18:26, 11 May 2008 (UTC)
I reread your request that WP:SPS be consulted. Since Watts is mostly publishing the surveys that other people are gathering, isn't this quite a bit different than somebody cooking up a theory and putting up a website on his own supporting it? Furthermore, wouldn't a professional weatherman be exactly the kind of person who would be considered an expert in the instrumental temperature record and therefore overcoming the normal burden of WP:SPS? Aren't these stations bread and butter for weathermen? TMLutas (talk) 19:18, 11 May 2008 (UTC)
No it wouldn't make it better - since you are both referencing self published surveys and original research based upon it. And no, a "weatherman" is not an expert on the instrumental temperature record. There is no need for any form of education to become a weatherman. And afaik Anthony Watts has no formal (or informal) education to make him an expert. If the research is worth-while, then it will end up in the peer-reviewed press, and then we can mention it. --Kim D. Petersen (talk) 20:33, 11 May 2008 (UTC)
Question for Kim. Is it not the case that realclimate.org is in fact a self published source? And is it not the fact that the wiki page for that site has a large number of self references?? SunSw0rd (talk) 19:56, 12 May 2008 (UTC)
Yep, RC is a WP:SPS - and it falls under the exceptions given in SPS. (Previously published experts talking about the subject that they are experts in) --Kim D. Petersen (talk) 00:10, 13 May 2008 (UTC)
You would think then that a weatherman with decades of experience and who runs his own weather equipment company would similarly fall under the same exception on a site that devoted to rating weather equipment. Do you agree then that surfacestations.org is just as meritorious as realclimate.org? TMLutas (talk) 22:35, 15 May 2008 (UTC)
You'd think you'd know the difference between weather and climate by now. Or between someone who makes the kit and the people who use it. Or... William M. Connolley (talk) 22:45, 15 May 2008 (UTC)
Please do not knee jerk your response. The types of errors on discussion (site placement bias, instrument bias, materials bias) affect both weather and climate equally. If latex paint ups temps on the stevenson screens, the bias does not only magically hit the weather community and not the climate community. 207.145.26.125 (talk) 17:10, 16 May 2008 (UTC)
And you should try not to knee-jerk your conclusions ;) No homogeneous bias has been shown in the temperature record yet. And a bias at regional locations may affect weather conclusions - but not necessarily climate conclusions. They are different beasts. --Kim D. Petersen (talk) 17:34, 16 May 2008 (UTC)
Whether or not the bias you describe has been demonstrated or not is irrelevant to whether a weatherman is qualified as the sort of expert that earns the exception in WP:SPS. These stations are dual use by both climatologists and weathermen. Both communities should be given the same sort of deference or neither of them should. If anything, surfacestations.org should earn wider latitude because its ambitions are much more objective. The reports are stored individually and anybody who would like to can challenge them/change them. Realclimate.org's narrative and debate format is inherently less subject to editorial oversight. That's not a criticism but rather an observation that they're not engaged in data gathering but debate and that's much more of a judgment call. So why does realclimate.org seem to get more latitude when it should be getting less? TMLutas (talk) 06:12, 26 May 2008 (UTC) (apologies, forgot to log in so I'm re-signing)
Just as many other large websites that are centered on a specific issues aren't mentioned on the respective issues. This one isn't either. If and when they publish something - or are significantly mentioned in the scientific media - then it should go in.
We as editors have no idea whether what they are doing is non-sense or a valuable contribution. And its not our job to decide it. That is reserved for secondary sources. --Kim D. Petersen (talk) 19:40, 3 May 2008 (UTC)
If you can provide a reasonable definition for significantly mentioned and scientific media (or references, that's fine too) around which to form a consensus, I think I can live with that. TMLutas (talk) 04:58, 7 May 2008 (UTC)

[edit] request: definition of global average temperature and how it is calculated

Hi there. The global warming article talks about average temperature of the Earth's near-surface air and links to this article here. However, I do not find any information here about how this average temperature is defined and calculated from the individual measured values. Also, the physical meaning of an average of temperature values is not inherently obvious. Check out this paper here [6] for more information. (Note that the article I'm referencing to links "global temperature" to the article on Climate, where I cannot find any useful information about the basics behind a global average temperature either). Am I right with suggesting that more information about that should be added into this article here, or rather into the articles about global warming, or climate? A substantial part of the global warming discussion relies on the concept of a global average temperature, but I've found no justification for that concept so far. FeelFreeToBe (talk) 14:25, 18 May 2008 (UTC)

A global average temperature exists, and is defined as... the global average temperature. But what you want is how the values are caculated. Wiki doesn't seem to have this info. I have no refs to had (except [7]). In essence... each station in converted into anomalies; these are then averaged into some suitable division, such as 5x5 degree boxes, and the global mean computed. You need some scheme for filling in holes, especially in earlier years. The sea is treated slightly differently; I'm fairly sure the recent Hadley datasets use some kind of EOF-guided interpolation to fill in the blanks William M. Connolley (talk) 19:26, 18 May 2008 (UTC)
If you follow the article's link 2, 3 and 4 the calculation methods and procedures are explained. It seems sufficient so. Gabriel Kielland (talk) 22:23, 18 May 2008 (UTC)
Even in the refs you're pointing me to I cannot find sufficient information. Sure, I think including something about how data is computed (division into grid, filling blanks, ...) would also do the article good. But I'm actually looking for the fundamental definition and physics behind a global average of temperature values. The thing is, that the use of such a value in climatology somehow implies that the "global average temperature" (as commonly used) is regarded as a meaningful physical quantity. As such, it requires to be defined in a physically unique way. Defining the "global average temperature" as an average is not unique because there are many ways to calculate an average. I'm disappointed that none of the refs I've read so far even points out which average is used. I think it's the arithmetic mean, but I'm still looking for the proof. Also I've seen some notions that some kind of a weighting scheme is used, but what are the values weighted with? For example from a physical perspective, it would make some sense to weigh the values with the local heat capacity of the air (which depends on pressure and temperature). Therefore a (sourced) mention of which average is used is absolutely necessary. Further I would suggest, since considerable doubt exists about the physical relevance of an average of temperature values (see the paper I referenced to above), a subsection about the physics behind the ominous "average global temperature" wouldn't hurt either, though it seems to be tough to find information about that. FeelFreeToBe (talk) 07:39, 19 May 2008 (UTC)
Disputing the existence of a meaningful global avg T in theory is the sign of a wacko who has been reading too much McKitrick and Essex. So is wondering if the avg is arithmetic or otherwise. Sorry if thats a bit harsh, but thats reality for you. The answer is that (of course) the averaging is arithmetic. No-one but the wackos will suggest otherwise William M. Connolley (talk) 18:06, 19 May 2008 (UTC)
I'm not sure how much of the following belongs on this discussion page, but your notion of me being a wacko is a bit too much to be ignored. Sorry if the following may sound a bit harsh sometimes, but it's primarily intended to be accurate and clear. First of all, I am NOT making any claims that I cannot defend here. Check out the WP article on average to find out that the word "average" alone does NOT automatically mean arithmetic mean. I AM reading the Essex/McKitrick/Andresen paper (and it's the first global warming sceptics paper that I'm reading) and I'm reading it critically and I'm checking the science VERY carefully. I've already worked my way through most of it, and so far I DO understand the science they're employing and I find that it's undeniably CORRECT. So are their fundamental conclusions (note that they do NOT make any judgment about whether global warming is happening or not, please read and understand their paper first to realize what they are in fact pointing out). I DO have some issues with that paper, but so far those are minor and have nothing to do with the underlying science or the basic conclusions. I HAVE SEEN that the authors are torn apart in the blogosphere, I NOTICED the "flaws" in their example calculations, but I also noticed that those are minor flaws that can be easily corrected and the example modified in a minor way (by changing one of the starting values) so that qualitatively the same results are obtained. I DO NOT CARE who the authors are or whether global warming is scientifically proven or not, I am just arguing on the basis of fundamental science. BTW, I do NOT believe that exchanging the arithmetic mean by another will SIGNIFICANTLY alter the results or the conclusions. But I DO believe that most scientists who are working with the "average global temperature" are NOT aware that this value is actually not a physical quantity (or has not been proven to be so), but merely a statistical index which is USED as if it HAD a physical meaning (which to my knowledge has not been proven). I am MERELY pointing out that the provided definition of an "average global temperature" is NOT sufficient (because it's not unique), and that no physical background for such a value is provided. I am currently NOT suggesting that Essex's/McKitrick's/Andresen's take on the subject should be used for expanding the article, because they are apparently not representative for the field, and what the article needs is a description of how it IS done (well-sourced, please), and the physical justification by those who ARE working with the "global average temperature". I cannot prevent you from not taking me serious, but I strongly encourage you to come up with the proof that the science and the fundamental conclusions of the author's or my own argumentation are incorrect, if you want to sustain your assertions. If you want reality, however, I can post a derivation here or on my talk page, that will show how the construction of a physically plausible definition of a global temperature can lead to either the arithmetic mean, harmonic mean, or more complex structures, depending on which non-trivial assumptions are made in the model. Let me know if there's need for further elaboration. FeelFreeToBe (talk) 22:58, 19 May 2008 (UTC)
Hm, the silence here disturbs me a bit. If you believe that this is just hairsplitting or that, for some reason, using the arithmetic mean is the only obvious, logical and reasonable choice, then I'm afraid you're fatally mistaken. Check out Molar mass distribution to see an actual physically important example of how different averages of molar masses in the case of a molar mass distribution of a polymer have a different physical meaning (and those meanings have been proven by theory!), or check out this WP article here [8] to learn about a specific case of physically reasonable temperature averaging. FeelFreeToBe (talk) 11:32, 21 May 2008 (UTC)

Offline for a bit. First, Essex and McKitrick are the wackos, not you. You are suffering from reading too much E&McK, which is bad for you. As is SHOUTING. All your examples are not using geometric, say, means, just weighted arithmetic. The T record uses weighting, of course, since a 5x5 cell at the equator is bigger than one at the pole. As for why E&McK are wackos, try http://rabett.blogspot.com/2007/03/open-book-test-in-comments-over-at.html William M. Connolley (talk) 16:57, 21 May 2008 (UTC)

First of all, apologies for the impatience and the capital letters. It was really not intended as shouting at all, I was just trying to make my statements as clear as possible. I understand this reads like shouting, so I won't do it again, thank you. As for the blog-reference: I've already looked at too many of those. I've looked into this one now, but nothing has caught my eye that would prove E&McK&A wrong. If you have a specific point, please let me know. But be careful: A vast majority of the contributors to these blogs don't really understand the underlying physics, as becomes obvious in some of the remarks from time to time. However, most reviewers of articles in a journal like non-equilibrium thermodynamis do know the physics, and to be fair, the physics that E&McK&A employ to prove their points is really not higher than undergraduate level. I won't judge whether the authors are wackos or not, but the science in their publication is undeniable and correct. If you know something about physical chemistry it would actually do you good to read it.
However, I realize that a discussion about whether they are right or wrong on this page might last forever (at least as long as you don't read and understand the paper), so most likely it won't lead to any improvements in the article. As I said, I'd prefer the information from the actual guys who are working with that stuff. The data treatment and averaging is far from trivial and at least it should be made clear in the article that this is so. I think useful sources could be the following: [9], [10], [11] and references given therein. Detailed descriptions of the data treatment can be found (though it gets arbitrarily complicated), but none of the papers I've looked at reasons a physical relevance of either the global average T, or the anomaly value (BTW, the "temperature anomaly" is used in the charts, but is not explained in the text. It's not a trivial measure, and is not the same as the average global temperature. If I find time, I might add something about it next week).
As to the examples I was giving, among the molar mass means Mw and Mz are weighed with the molar mass itself or its square (not sure if that's still arithmetic strictly speaking), Mv however has a power mean structure. For another example, look here to learn that joining ohmic resistors in parallel fashion leads to a resistor with the harmonic mean of all the individual resistors. Another example I can give you (can't source it unfortunately) is again the temperature after joining two systems. You point out yourself that the individual temperatures are weighed with the size (proportional to the volume) of their gridboxes. This may seem trivial or obvious to you, but if you think about it, there's nothing else behind that thought than to scale the temperatures according to the total number of gas molecules in that box. The physical aspect behind that is that the internal energy is proportional to temperature times number of gas molecules. By assuming that the number of gas molecules only depends on the size of the box, you arrive at your weighed arithmetic mean. However, the number of gas molecules also depends on pressure and temperature (according to the ideal gas law), therefore you would also need to include those in the weighing scheme. If you work out the equations, you will in fact end up with a weighed harmonic mean instead of an arithmetic mean (because the number of gas molecules scales with 1 devided by T)! In fact just today I tried to treat the original gridbox data (from [12], I chose just the years 1997 to 2007) with both models and see what happens (lacking pressure data, I assumed the pressure to be the same for all boxes in the second model, which is a non-trivial assumption as well). As I expected (see above), the differences between the two models are not significant at all (the anomaly value of both models is in very good agreement), though indeed a few month can be found in which the one model shows a warming trend, and the other a cooling trend or the other way around (but only for a single month in a row). However, the complete picture remains unchanged. My point is simply this: E&McK&A are right, but personally I don't expect a big impact from their findings. I can just hope that finally Climate scientists will equip their "global average temperature" and "temperature anomaly" with some fundamental science and a real physical meaning in theory. FeelFreeToBe (talk) 23:46, 21 May 2008 (UTC)

there's nothing else behind that thought than to scale the temperatures according to the total number of gas molecules in that box - no; its according to the area of the planet being represented William M. Connolley (talk) 07:02, 22 May 2008 (UTC)

My argument is that in order to design a physically meaningful global temperature (and my particular example is in fact to define the global temperature as the final temperature after equilibration through a certain thermodynamical process) you have to weigh the values with the gridbox size because the heat capacity (which is proportional to the number of gas molecules) is proportional to the gridbox area, and out of no other reason. But apparently what is done is to ignore physics and just weigh the values with the area because it seems the right thing to do - again, without any physical background. Therefore, the global temperature which is used is physically absolutely meaningless. To claim that this value can indicate whether an alleged physical process like "global warming" is taking place would simply be an interpretation of that temperature index without any foundation.
Let me try to give you a demonstrative example. Analogies to the global warming case are given in parenthesis. Consider a storehouse (globe) that has both cotton (cold areas) and silk (hot areas). The price of cotton is 2$/m2 (temperature of cold areas), the price of silk is 20$/m2 (temperature of warm areas). Now all the pieces of textiles are stored in paper boxes of equal size (surface area grid boxes), and the number and size of the boxes never change. However, the amount of textiles (number of gas molecules, total heat capacity of the air) in each of the boxes may change due to storage, withdrawel, or re-sort (climate dynamics, local warming or cooling). Let's assume for simplicity that we have such a large number of boxes that there's either cotton or silk in each box, but never both of them (homogeneous temperature within each gridbox; only two possible temperatures for each gridbox in this particular example). Now we are interested in the "storehousal average price" ("global average temperature") in order to find out whether "storehousal cottoning" (global cooling) or "storehousal silking" (global warming) is occuring. One possibility to form an average would be to simply calculate the arithmetic mean of the textile prices of all boxes, an easy task, given that all boxes have equal size. Another possibility would be to weigh each price value of a box with the amount of textiles in that box (take the variable density of air into account). Now let's calculate those averages for a particular case: Consider 4 boxes with the following content: box 1: 1m2 cotton, box 2: 1m2 cotton, box 3: 0.2m2 silk, box 4: 0.2m2 silk. Try to do the calculation yourself (the prices of cotton and silk are given above)! The results for the "averages" in question are (with "average#1" being the former and "average#2" the latter): average#1 = 11$/m2, average#2 = 5$/m2. Now we store 0.6m2 silk into the last box and also put the 0.2m2 silk from the third box into the last box, while cutting the cotton from box 2 in two pieces of 0.5m2 each, putting one of them into the third box: box 1: 1m2 cotton, box 2: 0.5m2 cotton, box 3: 0.5m2 cotton, box 4: 1m2 silk. Now the "averages" change to: average#1 = 6.5$/m2, average#2 = 8$/m2. The result is that when we use average#1, we observe "storehousal cottoning" by 4.5$/m2, while when using average#2, we see "storehousal silking" by 3$/m2. If you don't care for the physics, then there is absolutely no way to decide between the two averages. If you do care for the physics, then please read the E&McK&A paper. FeelFreeToBe (talk) 12:48, 24 May 2008 (UTC)
This is all meanignless, because the heat stored in the air is fairly small, so you're not interested in its heat capacity William M. Connolley (talk) 13:10, 24 May 2008 (UTC)
You're totally missing the point. The absolute value of the heat capacity is of no importance for the arguments given above. The only thing that counts is the relative difference between the heat capacities of different places. If all heat capacities are equal, then they cancel out in the weighing scheme - not because they are small, but because they are equal. If they are different you can express them as some "standard" value times a relative deviation from that standard. Then the standard value cancels out and you're left only with those relative deviations.
If your point is that the heat stored in the air is meaningless for climate, then the question is rather how much significance an average of those air temperatures has at all. However, this problem is a whole different story and has absolutely nothing to do with the question how to do the averaging. FeelFreeToBe (talk) 14:01, 24 May 2008 (UTC)
The temperature is essentially a skin temperature. It doesn't relate to heat capacity. Its used because those are the measurements we have. As for E&McK, you must have missed [13]. But I'm sure you'll explain it away William M. Connolley (talk) 15:59, 25 May 2008 (UTC)

If you would take your time to read and understand my arguments and the E&McK&A paper you would realize that an arbitrary average of temperatures is not automatically a temperature itself, and that the average which is used has not been derived by physics at all. Therefore your "skin temperature" is not a temperature itself, but rather a statistical index (with the unit of a temperature) that may tell you something useful, but not necessarily does so. To fully understand this point you really need to read and understand E&McK&A's paper (at least the first part of it), or the textile storehouse example I was giving above (in that example the average#1 price per m2 doesn't have the "physical" meaning of a price any more). Simply defining an average as "skin temperature" has no physical foundation and no physical implication by itself. Those claims remain to be proven. The heat capacity comes into play in my own personal attempt to design a physically meaningful average. I'm not claiming that my solution is any better than the current one. It just serves as an example to show a possibility of how to average temperatures on a physical basis.
I believe I have provided you with sufficient arguments and good examples to help you to understand the point, but obviously you're either incredibly stubborn or you simply didn't pay attention to the arguments or you don't understand the issue. I for my part have always replied to your claims (which are really nothing more than unsubstantiated claims), using plausible arguments and by explaining them with undeniable physics, not by explaining them away. These solid arguments are nothing that you can just slur over, but that's exactly what you did.
As for the Tim Lambert Blog, I read that one some time ago. I have by no means made any claims against my own knowledge here. It's true that it's embarrassing for scientists to make such silly mistakes, but the actual science they are doing remains untouched. Just because it's possible that two different averages can show different trends doesn't mean that they will show different trends in every example. Above I said myself: "I do NOT believe that exchanging the arithmetic mean by another will SIGNIFICANTLY alter the results or the conclusions." But I've also provided you with examples that do prove E&McK&A's point, in addition they are providing more examples in their paper (note, however, that one of them (coffee cup example) has another flaw, but by changing one of the (arbitrary) starting values it works even after correcting that minor mistake). Any credible scientist would immediately notice that those minor flaws have absolutely no impact on the correctness of the scientific points and conclusions of E&McK&A. The fact that Tim Lambert obviously ignores this speaks volumes. However, if you have no knowledge in physical science yourself, and if you trust Tim Lambert's skill more than mine, then why don't you invite him do have a look at what I wrote here and reply to it? If you have any reasonable doubt that my logic is wrong, please point me to the mistakes that you think I'm doing. FeelFreeToBe (talk) 01:31, 26 May 2008 (UTC)

Your argument is defensible, but it's not the one offered by Essex&Co. Physics does tell us how to average temperatures, just as it does tell us how to average fabric prices in your example. You've shown that Essex&Co are wrong. You're right that the conventional method (weight by cell area) is a simplified version of the correct physical model. I doubt that this makes a significant difference to temperature trends -- if it did Exxex&Co would have told us instead of the contrived examples they give (the average of the 200th power of the temperature!!). --TimLambert (talk) 08:44, 26 May 2008 (UTC)
WMC is an actor in the debate on global warming. He is an advocate and works in the field. You might profit from googling his name to understand who you are talking to. What conclusions you draw after that are up to you. TMLutas (talk) 06:12, 26 May 2008 (UTC) (apologies, forgot to log in so I'm re-signing)
Thanks for that clue, TMLutas! FeelFreeToBe (talk) 22:05, 26 May 2008 (UTC)
Tim, thanks for your quick respose! I'm glad that we already agree on a few things (for example your last sentence). However, there is still need for further discussion. Let me reply to some of your statements:
Your argument is defensible, but it's not the one offered by Essex&Co. For most people, simply taking the arithmetic mean and calling it average temperature just seems to be the obvious and only logical choice (most people just don't care about the physics), which makes it hard for me to convince people who have apparently fallen into that trap. That's why I needed to come up with an alternative average that can be demonstrated to make even more sense than the beloved one. This is not a point that Essex&Co are interested in making in their publication (see below why).
Physics does tell us how to average temperatures. That's wrong. Physics does not tell us how to average temperatures, physics can only tell us how to calculate the temperature of a homogeneous system or subsystem, or multiple temperature values in form of a scalar field in the case of inhomogeneous, non-equilibrium systems. If we decide to calculate the temperature that the atmosphere would have after a well-defined physical equilibration process, physics can tell us how to do that. But that temperature depends on the process that we choose. The process I chose in my model is the following: Keep the total internal energy constant; consider all gridboxes as closed containers with fixed volume; allow the temperatures to adjust to an equilibrium value; the pressures may vary individually in each box. There is an infinit number of possible processes that you can choose from. For example, if you chose a similar but reversible process (entropy is constant) by allowing work to be extracted from the gas, the final temperature is the (weighed) geometric mean instead of the arithmetic.
You've shown that Essex&Co are wrong. No, actually I haven't, nor have you or anybody else. In order to understand that my model is by no means a simple solution to the problems that Essex&Co are revealing (that is lack of physical basis), one has to question whether my model is fulfilling the requirement to be physically meaningful. Being the (real, physical) temperature of a hypothized system after a certain equilibration step is only one part of the story. The missing part is: What physical relevance does that value have for the global climate? I cannot answer this question, nor can Essex&Co, which is most probably why they don't propose any alternative model themselves. What they say is this: In special circumstances averaging might approximate the equilibrium temperature after mixing, but this is irrelevant to the analysis of an out-of-equilibrium case like the Earth’s climate. Just for clarification, consider an example in which this problem doesn't occur: The number average molecular weight of a polymer is proven to be physically meaningful: If this particular average value increases then a decrease of the osmotic pressure of a solution follows strictly (keeping the total mass constant). However, to my knowledge, such a physical connection between the "average global temperature" and physical climate processes has neither been proven nor demonstrated.
You're right that the conventional method (weight by cell area) is a simplified version of the correct physical model. If you know any reliable source for that "correct" physical model and its simplifications, please let me know. That's exactly what I'm searching for since the beginning of this discussion. That's what is needed for improving the article. Regards. FeelFreeToBe (talk) 22:05, 26 May 2008 (UTC)
This paper here might be of some use: [14]. Unfortunately it's rather old and is also not well-cited in the literature of the scientific community. I'm still searching for better references. If anyone here finds a paper that clearly defines the currently used "global temperature anomaly", possibly with some physical background as well, please post the reference here! Thanks. FeelFreeToBe (talk) 08:23, 1 June 2008 (UTC)
Here's another interesting paper (Pielke et al, 2007), concluding: This paper presents reasons why the surface temperature is inadequate to determine changes in the heat content of the Earth’s climate system. Interestingly, it employs a non-trivial definition of global average surface temperature based on heat content of the land-ocean-atmosphere system, radiative forcing, and climate feedback. That definition is obviously taken from a 2005 NRC report. FeelFreeToBe (talk) 09:18, 5 June 2008 (UTC)
If you liked that, you'll find more of the same on his blog; its a line he's been pushing. What he is "forgetting" is that the ocean heat content isn't well enough known to construct an accurate record from William M. Connolley (talk) 18:25, 5 June 2008 (UTC)
Thanks! So, let me sum up: We have one method to calculate an apparently meaningful quantity, but it's not possible to calculate it accurately (that is, with small enough error bars in order to conclude something); and we have another (commonly used) method that provides an accurate record, but in order to conclude something from it you first have to believe that it is physically meaningful... Splendid! FeelFreeToBe (talk) 07:05, 10 June 2008 (UTC)

Ok, sorry for the fruitless discussion above (in terms of improving the article), but I think there's one thing at least that should be done, and that's adding some information about the temperature "anomaly". If I have understood correctly, for each measurement station, averages of absolute temperatures are calculated for a reference/base period (1961-1990) for each of the 12 month. Those averages are then substracted from each individual temperature value (after calculating monthly averages) to yield the "anomaly" value for that station for that month of a particular year. Those values are then used for spatial averaging (to obtain the gridded values, hemispheric averages, and global average), and further temporal averaging (to obtain annual averages, or to smooth the plots of glob. temp. anom. vs time). Now, the reason for using the anomaly values instead of absolute temperatures is that a number of possible systematic errors cancel out in the substraction step.[15] Do you know any reference which comprehensively describes all the advantages of the anomaly values over absolute values? I think it's necessary to provide that information, primarily out of the following reason: The "surface temperature record" plot shows monthly averages. If not pointed out that reference values are substracted for each month, and those reference values are different from each other for each month by several °C (!!!), this plot is misleading the reader, since the plot that one would expect to be looking upon actually would show huge annual oscillations without an obvious trend. (BTW, I suspect that those oscillations come from the fact that there's much more land and less ocean on the northern hemisphere, but it would be nice to see the reference in which those discrepancies have been identified and found to be taken care of by using anomalies). I'd be happy if you could point me to the sources. Thanks. FeelFreeToBe (talk) 07:05, 10 June 2008 (UTC)