I see the thread has moved quite a bit since my last post - I apologize if this disrupts the flow, but I’ll address where I left off and after the holidays read the rest of the thread to catch up.
(in response to my comment about moving averages being lagging indicators and longer moving averages becoming less sensitive to trends)
Where n is the number of time periods. When n grows larger, α grows smaller, and the moving average more closely follows the past than the present.
In either case, as n increases, the moving average tends to more sluggishly respond to trends, regardless of the direction of the trend. We can imagine three trend directions:
[ul]
[li]Increasing trend – a larger n will tend to make the moving average lower and it will lag behind the trend more than with a smaller n.[/li][li]Decreasing trend – a larger n will tend to make the moving average higher because there are more past terms (y[sub]t-n[/sub] for example) that are large relative to present values.[/li][li]Steady or flat trend – a larger n will tend to reflect values from before y became flat – higher if y[sub]t-n[/sub] and its fellows are large, lower if y[sub]t-n[/sub] and its fellows are small.[/li][/ul]
In none of these cases does the moving average more rapidly follow a trend when n is higher or become more sensitive to trends when n is higher.
A 40-year moving average may be appropriate when you’re talking about 1000 years of proxy temperatures, but it is inappropriate with 100 or 150 years of instrumental temperature data.
If we disagree on the mathematics above, I don’t see much hope of us being on the same page.
Generally the longer moving average will always be smoother - but remember that the moving average is a trend analysis tool it does not replace the data. Like every trend analysis tool, it can be misleading if used improperly - such as a 40 year smooth on a 100 or 150 year temperature record.
You’re misconstruing the analogy - instrumental temperature records have a far higher resolution than any of the proxies, in temperature precision, global reach, and temporal accuracy. They show fundamentally more accurate data, and can be examined at a different scale than the proxy record can.
Unless you can see a trend in the recent temperatures (the last 1000 years does include the last 150 years) there’s little point in comparing a non-existent current trend to any trend further in the past.
Well, I’m off to Mexico for two weeks - I wish you all a Happy Holiday and hope you and your loved ones are well. See you in the new year!
This is the key point that I’ve been making. The fact that you look at the instrumental record alone and see an “elbow” (actually multiple elbows) doesn’t necessarily mean anything and doesn’t undercut my basic point about splicing.
That’s almost certainly true. And it shows why the hockey stick is wrong. If the proxies are not fully capturing the temperature trends of today, it follows that they do not fully capture the temperature trends of 1000 years ago.
And if you CAN see a trend in recent temperatures, you need to ask yourself if there is any way of measuring similar trends in past temperatures. Conversely, if you find a trend in past temperatures, you need to ask what happens if you apply the same technique to recent temperatures.
Do you contest that the first study I cited shows what I said it said, since you’re posting no criticism of it?
Let’s address the first study first.
This is in direct contradiction to any claim that the records diverge. This warming is in line with, and perhaps a little greater than, what we observe with the instrumental records.
But you’ve also argued that the temperatures diverge in the 20th century - you cannot claim both that the proxy record is wrong because it diverges, and then claim that the divergent part proves the proxy record.
But let’s settle the factual issues first - what are the slopes in the picture you’ve shown? Saving as a JPG, and clipping out the relevant parts, I find the the time period from 1850 to 2000 is 114 pixels wide and 61 pixels tall (slope = 0.54). The slope from 1000 to 1150 is 117x57 (slope = -0.50), from 1150-1300 is 114x36 (slope = -0.30), from 1300-1450 is 117x6 (slope = +0.05), 1450-1600 is 116x17 (slope = -0.15), from 1600-1750 is 112x39 (slope = +0.35), and 1750-1900 is 114x16 (slope = +0.14). So the simple claim that there’s no dramatic change in the slope is simply false altogether.
False, as shown above.
The height of a 1[sup]o[/sup]C anomaly from the picture you linked is 85 pixels tall, which would suggest that the warming from 1850-2000 is 61/85*1 = +0.72[sup]o[/sup]C.
The same analysis on File:Instrumental Temperature Record.png - Wikipedia gives a height of 309 pixels for 1853-2000 with an scale of 260 pixels per 0.6 degrees, or an instrumental temperature record of 260/309*0.6 = +0.71[sup]o[/sup]C. I suppose you could criticize the fact that the proxy data doesn’t go all the way until 2000, but that would again only indicate that the proxy temperature changes are greater than, rather than equal to, the instrumental temperature record.
No. The bottom line is that you have made a vast number of unsubstantiated claims. The records don’t diverge at all.
For a variety of reasons, brazil84 has rejected each of the other data sets. At first, I wanted to compare the instrumental temperature record with proxies - he claimed proxies were worthless. He then challenged the instrumental temperature record and claimed the urban heat islands were the cause. The only thing remaining is satellite data, which only goes from 1980-2000.
Do you have any raw data, so I don’t have to cut-and-paste and count the pixels as I have done above?
That there is no difference between the two trends is a point in my favor. I am using the satellite data to show that the instrumental temperature record is reliable, and the instrumental temperature data to show that the proxy data is reliable.
For the purposes of my argument, I’ll gladly accept “barely significant”. Brazil84 claims there’s no evidence the Earth has warmed at all; a small, but non-zero increase is sufficient to disprove that.
I am unqualified to support or contest that statement - certainly people who publish in Science think there’s a relation, and their reviewers did. Is there an equivalent amount of peer-reviewed literature on the other side?
The abstract of the paper I quoted above claims there is good agreement between “global” data sets. Would you care to explain the discrepancy? (e.g. Are there glaciers on Kilimanjaro that he might be using that other people contest?)
I request peer-reviewed articles to counter peer-reviewed articles. I don’t cite RealClimate.org as an authority in favor of my arguments; I’d appreciate it if others wouldn’t cite equivalent organizations in favor of theirs.
On a very broad scale, I agree that temperature proxies don’t “prove” anything. But then again, I created a whole other thread for proving global warming without citing any temperature proxies at all. You are welcome to participate. (brazil84’s Global Warming Thread).
I’m fine with people questioning the temperature proxy record. However, you cannot both attack the temperature proxy record and claim anything about a Medival Warming Period or Little Ice Age. If you want to say the proxies are shit, that’s fine - in subsequent discussion, just don’t ask why the Earth was warmer in the MWP. Either you accept the proxies as a whole or reject them as a whole.
I fear that this quote perfectly exemplifies and encompasses the problem with your worldview. Proxies are like anything else, good, bad, and indifferent. Each one has its own unique problems and strengths, both as a class (e.g. tree rings) and individually (e.g. Thompson’s widely cited but unarchived data). Beer intake is a good proxy for level of intoxication, while age is a poor proxy for intelligence.
In addition, it is quite possible to get bad results from any proxy, valid or not, by making any one of a number of statistical errors and mis-steps (e.g. Mann’s Hockeystick).
Thus, it is obviously necessary to look at, not only each individual proxy, but at each individual proxy-based study, and to judge it on its own merite.
You, however, have a black-and-white view (accept them all or reject them all) which has no relationship to science.
w.
PS - I had said
To which you replied
Umm … the HadCRUT3 and GISS temperature datasets are raw data … are you sure you’re familiar with the field of science we’re discussing?
Slope from beginning of graph to Year 1000: 0.54
Slope from year 1000 to year 1050: -1.04
Slope from year 1950 to end of graph: 0.12
In any event, try to imagine the first derivative of that graph.
My point stands.
Esper’s graph shows a temperature change of +0.05C between 1950 and 1992 (the end of the graph).
Are you confident that the instrumental record shows a similar change?
Of course I do - it doesn’t even cover the time period you claim it does.
Straw man. I’m not trying to “prove” the proxy record. My point is that if one accepts YOUR proxy record, it doesn’t show what you claim to be the case – a dramatic increase in rates of change of temperatures.
The only way to show your claim is to accept the proxy up until 1900 or so, and then reject it in favor of the instrumental record.
I do not have the expertise to look at each individual proxy.
(1) Do you?
(1a) If you do, which datasets are acceptable and which are unacceptable?
(1b) Are there any datasets that you consider acceptable that show both the existence of a MWP and LIA and temperature divergence in the 20th century? Which one(s)?
(2) Both authors and reviewers at Science magazine accept these proxies.
I was asking for the raw data in a chart, rather than graph, format. That would allow a quick calculation of warming and cooling rates without having to use the Count The Pixels method of temperature estimation.
Are you sure you understood the question before answering?
I don’t contest that you can cherry-pick. I kept each of the time frames constant in 150 year chunks, and I evaluated each of the 150 year chunks throughout the graph. You did neither.
You did not argue that the temperatures only diverged in the “last” 42 years - you claimed they diverged for the instrumental temperature record, which is 150 years. I took 150 year chunks throughout the graph and showed that the increases neither diverge nor are they large compared to recent warming.
You have taken smaller chunks for random ranges and compared them with mis-sized chunks.
Verifiably false.
[GEOGRAPHIC REGION: Global
PERIOD OF RECORD: 1600-1990 AD
aptronym, I took a look at the glacier reconstruction you cited (Science 29 April 2005: Vol. 308. no. 5722, pp. 675 - 677), and was less than overwhelmed …
They claim to be able to tell the global temperature in the year 1600 with an error of a quarter of a degree.
Now, in the year 1600, they have exactly three glacier length data points, two of which are only about 100 km apart, and all of which are in Northern Europe.
The current error for instrumental data (HadCRUT3) with hundreds of stations worldwide is estimated by the HadCRUT folks at about an eighth of a degree.
Now tell me, aptronym … if we can only tell modern global temperature to an eighth of a degree or so, do you really believe that the length of three European glaciers four hundred years ago can tell us the global temperature at that time to a quarter of a degree? Just three glaciers?
Because if so … I’ve got a bridge for sale that you might be interested in …
Plus, if what they say is true, we could get rid of all of our modern temperature stations, and replace them with a couple dozen glaciers, it would be much cheaper.
Or, if just three European glaciers can tell the global mean temperature to a quarter of a degree, we could just replace them with three temperature stations (which have to be more accurate than glaciers) and we’d know the world temperature to better than a quarter of a degree.
So while you, and the author, and the reviewers might believe that study … I don’t. It doesn’t even pass the smell test. Three European glaciers four hundred years ago can tell the global temperature to a quarter degree? Riiight …
This is a perfect example of why peer review can’t be trusted to do anything than remove the most obvious of errors.
You are free to be underwhelmed or overwhelmed, but unless you have some particular expertise in climatology, it’s really irrelevant.
(1) Do you?
(1a) If you do, which datasets are acceptable and which are unacceptable?
(1b) Are there any datasets that you consider acceptable that show both the existence of a MWP and LIA and temperature divergence in the 20th century? Which one(s)?
(2) Both authors and reviewers at Science magazine have accepted this as an acceptable proxy.
This is an applaudable evasion of the issues at hand. The issue of whether the proxy data is representative of global temperatures is an issue that is completely separate from what we have discussed so far.
The question is whether the proxy data matches the instrumental data in the overlapping time frame.
(1) If YES, then the issue is settled - brazil84 claims to reject all proxies based on divergence, and if there is no divergence, he has to accept them.
(2) If MAYBE, then the issue is again settled - brazil84 is making a positive claim that peer-reviewed, published scientific data should be ignored. If there is any ambiguity, again, I appeal to the fact that authors and reviewers in Science have accepted such proxies. In the absence of any authoritative reason why they should be ignored, they should be accepted.
(3) If NO, then I have one final question for both you and brazil84 - if the proxies irreconcilably diverge (i.e. even if you show a divergence I would argue that it may not be irreconcilable), then how would you ever know there was a Medival Warming Period or Little Ice Age? Without a proxy record, no temperature statements can be made prior to 1850 at all. Our understanding of a MWP and LIA come from proxy records - if you reject proxy records, you cannot then claim that there was even a MWP.
(quoting again so I can answer your question proper)
I believe that multiple glaciers, plus multiple tree-ring chronologies, all paint a consistent picture across most of the globe. That the data is not global per se is a caveat when looking at a single study, not a reason for rejection of all data.
The single study itself does not prove global temperatures - multiple, independent studies (glaciers, tree rings, sediments, etc.) which all show similar trends proves global temperature shifts.
In other words, you are taking a position which is directly contrary to peer-reviewed science. I want you to be absolutely clear, because this is a fundamental shift from your past argument - you’re going from “science supports my position” to “science is wrong.”
So you no longer dispute my claim about the slope of the graph?
Simple yes or no question. (Not that I expect an answer.)
Let’s take 1950 to 1990, which encompasses the bulk of the “back half of the century.” According to your own calculation, the proxy picks up well under half of the warming as allegedly detected by instruments.
It depends what you mean by “reject” and “accept.” But anyway, I mean exactly what I say - nothing more, and nothing less.
Assuming for the sake of argument that the proxies irreconcilably diverge (and I don’t know this to be the case), one can still be pretty confident that there was a medieval warm period based on the proxy record. You just don’t know the temperatures with a lot of precision.
For example, during the medieval warm period, there was apparently agriculture in parts of Greenland that are now subject to permafrost. If you combine this with similar types of evidence, you can infer that things were probably warmer then than now.
No, it looks like you’ve arbitrarily picked a random date which happens to coincide with your prejudice.
1981-1990 shows a warming greater than the instrumental record. How come you don’t cite that as your example?
Your claim was that the proxies diverge from current data. Picking a certain set of numbers and a short time frame only plays with the short-term variation on temperature.
For example, I could move back a year and make the proxies look worse …
Time Frame: Treering Instrument
1950-1990: +0.168[sup]o[/sup]C +0.587[sup]o[/sup]C
… or forward one and make them look better.
Time Frame: Treering Instrument
1952-1990: +0.184[sup]o[/sup]C +0.336[sup]o[/sup]C
I could point to the period from 1858-1990, where both warmings are exactly +0.777[sup]o[/sup]C. Hey, look, a perfect match; am I right? No, that proves my point no more than the 1951-1990 data proves yours.
This is why I advocate taking as long a time as possible - average as many 11-year solar cycles; smooth out as many anomalies due to volcanic eruptions; etc. Any way you cut the data, you see a modest warming from the 1850’s to the 1940’s; a small cooling from the 1940’s to the 1960’s, and then another modest increase from the 1970’s to the 1980’s in both data sets. The longer of a time period you look at, the more reliable the results are. Data based on short time periods are simply artifacts of noise.
My analysis does not depend on the exact date picked. Yours does.
The “back half of the century” quote is yours, not mine.
Allow me to remind you of the quote that you contest.
You completely understand - when you filter out the year-to-year variation in temperatures, you can get a decent picture of the centennial-length climatological shifts.
Now simply apply the same statement to the entire proxy record. It’s not contested that the absolute temperatures are pretty nebulous - so nebulous, in fact, that scientific publications use “Temperature Anomaly” instead of “Kelvin”. Ice cores sometimes just report [sup]18[/sup]O/[sup]16[/sup]O ratios rather than the corresponding temperature. Few attempts to measure absolute temperature make it into the publishable realm.
That is why the only analyzable characteristic is temperature difference. While the absolute temperatures are debatable, the difference remains constant. You are free to pick a time frame between 100 and 150 years, and I will analyze the data to show that the long-term averages support my assertion.
By that measure, you can be sure that global warming is happening because the ice cap keeps shrinking. Poor polar bears stranded in the Arctic Ocean! Icebergs the size of Rhode Island breaking off of Antarctica!
Pul-eeze.
I don’t cite this crap in favor of global warming nor do I accept the same crap against it.
I forgot to single this out and it’ll probably take me more than 5 minutes so I’ll just put it in a new post.
That things were warmer then than now is completely irrelevant to the topic of discussion. I never claimed that we are at the warmest point ever; I claimed that the temperature now is increasing faster than it did before.
I don’t contest that humans can adapt to any absolute temperature - even +/- 10[sup]o[/sup]C. Even our cavemen ancestors were able to do it at the end of the last Ice Age - but that change happened over several hundred or thousands of years.
Thus, it’s the rate at which temperature changes which is the most relevant. That we can successfully plant crops in Nunavut if warming turns it into a breadbasket is not in doubt - whether we could do it in a decade without massive famines is. That Siberia and Canada would become the new world superpowers is no problem by itself - but if it were to happen in the space of a decade, we’d have some serious political problems.
The problems of global warming stem from the rate of temperature change, not the absolute temperature.
Too bad the only facts you’ve cited to dispute the claim were cherry-picked. As I suggested before, there’s no rule that says the slope of a graph must be evaluated in 150 year chunks.
All you’ve done on this point is obfuscate.
False. Anyway, since I’m the one questioning the proxy, it’s my privilege to pick out a time period. 1950 to 1992 (the end of the study) is a reasonable choice.
The proxy record you chose diverges, even though the study authors no doubt calibrated the proxy as best they could to the instrumental record.
The bottom line is very simple. Just look at the graph with your own eyes: There’s no hockey stick.
It’s as simple as that.
Not sure, but one can infer that things are likely warmer now than they were in the 1970s when the artic had a lot more ice.
Lol. Obfuscating again, are you? Do you remember that a couple posts ago you asked me and intention how one could know that there was a medieval warm period? I guess that’s “irrelevant” now. Lol.
Mexico was great – I hope everyone had at least as good a Hannukah/Christmas/Kwanzaa/Winter Solstice/Deforestation and Consumerism Festival time as I did.
Here’s why I don’t see it that way. There are two requirements for a “hockey stick” like figure to exist:
A long, relatively flat “handle.”
A shorter, steeper, “blade” at the end.
An “elbow,” as I mentioned in my first post in this thread, is a poorly defined feature that is a result – not a cause of - the two features listed above. “Elbows” cannot be used to define a hockey stick and are the least easily determined part of it, simply because they are subjective – one person’s “elbow” is not another person’s “elbow.” Which inflection point do you choose as your elbow? 1902? 1920? Ask different people and you’ll get different answers. However, you can identify a “hockey stick” like shape even if you don’t know which inflection point is the “correct elbow” – if such a thing as the “correct elbow” could even be said to exist.
What do I see when I look at the instrumental temperature record? A relatively steep “blade.”
What do I see when I look at the paleoclimate proxies? A curve that goes up and down, but generally less steep than the “blade” in the instrumental temperature record, making a “handle,” if you will (actually, I’m pretty sure you won’t .)
My point being that criticisms of a “hockey stick” shape based on the “elbow” are considerably less convincing than, say, an extremely warm Medieval Warm Period would be.
It’s very helpful that from about 1856-1890 you can see in the instrumental temperature record a relatively less steep portion which I would consider (and I’m guessing you would not consider) part of the “handle” of a “hockey stick.”
And you think paleoclimatologists do not ask these questions?
Have you read any of the portions of their work where they describe the calibration, validation, and reconstruction portions of creating a paleoclimatic reconstruction? Or do you believe the answers they have come up with in these portions are incorrect?
The Earth has gotten colder and warmer in the past millenia, centuries, decades, etc.
The Earth obviously has cyclical changes to it that are beyond our control and influence.
It’s probablt practical to assume that the more people and industry we have, the more pollution we have, which is harmful to the Earth and it’s land and waterways.
If we can reduce our output of emissions that affect these issues, then it’s good, right?
Do we need a politician or a government to tell us this? Can’t we figure out a way to slow this down on our own?
Indepedence on oil would be a GREAT start, both for the Earth AND for the foreign policies of Western nations.
The infrastructure of say the USA is designed for wheeled vehicles. Can we not manufacture wheeled vehicles to travel along the paved roads without being powered by internal combustion engines?
I know, I know, special interests. Defeated yet again we are.
(1) Yes.
(2) I am not clear what you are asking here. Often, the problem lies not with the datasets, but with the interpretation placed on them. In this case, the dataset seems reasonable enough … but claiming we can tell the temperature 400 years ago to a quarter of a degree with this dataset is nonsense.
(3) Dunno … if a proxy shows divergence in the 20th century, I tend not to pay attention.
Yes, that is true of everything that Science magazine publishes, it has been accepted by authors and reviewers. I truly, truly hope that you don’t think this means anything about the truth or falsity of a given paper.
Perhaps that was your question, but I was not attempting to answer that question. My question was whether the glacier study was believable. I found it not to be so. You have evaded the question of whether we should believe the claims made by the author of the glacier study.
There are a variety of proxies which do not have the “divergence problem”. However, that doesn’t mean that they are necessarily valid, or that the conclusions drawn by a particular author from a given proxy are correct.
You have not “answered my question proper”. You have not answered it at all. So let me ask it again. Do you really believe that the length of three European glaciers four hundred years ago can tell us the global temperature at that time to a quarter of a degree? It’s a simple question, a yes or no will do quite nicely, and anything else is obfuscation.
If YES, then we can throw out our thermometers, if three glaciers show global temperature to a quarter of a degree, then thirty of them would replace all of the world’s temperature stations at a fraction of the cost.
If NO, then why would we believe anything that the study’s author’s say?
And while you say that “The single study itself does not prove global temperatures”, the authors clearly claim that their single glacier study shows accurate global temperatures. Are you saying you don’t believe them?
Ah, I see the problem. You think that if a paper is published in a peer reviewed journal, that makes it “science”, and thus if I disagree with a peer reviewed paper, I am saying “science is wrong”.
There are two problems with your claim. The first is the ludicrous idea that peer review establishes scientific truth. It does nothing of the sort. In some fields of science, well over half of the peer reviewed papers are eventually shown to be wrong. Mann’s original “hockeystick” paper contained egregious errors … but it was peer reviewed.
Here is how science progresses. Someone writes a paper, and someone else tries to find fault with it. If they can find fault with it, the paper is rejected. If no one can find anything wrong with the paper, it is provisionally accepted as scientific truth. The problem is thinking that peer review establishes the truth. It does not. The value of a scientific paper is how well it stands up to examination, not by peer reviewers, but by other scientists.
ALL PEER REVIEW DOES IS FIND OBVIOUS ERRORS. It does not guarantee that the claims of the author are correct. It does not make that paper “science”, as you seem to think.
The second problem with your claim is the idea that ordinary mortals such as yourself must accept all peer reviewed papers without question. Dude, you have a brain. Use it! Think about the claims the authors are making. Are they reasonable? Do they make sense? Scientists are not gods, and many of their claims are demonstrably false to even an untrained eye.
This glacier paper is a perfect example. As I pointed out, they make the remarkable claim that we can determine global temperatures to within ± 0.25°C, four hundred years ago, from the length of three northern European glaciers …
Now perhaps you just nod your head and say “author said it, reviewers passed it, it must be gospel scientific truth” … but in my world that claim doesn’t even pass the smell test.
.)