Statistical Question- re: AGW

[casebro]

So nobody wants to discuss the "merging proxies"?

To re-iterate, based on the graph posted in the "Global Warming Swindle" thread, poist #204, the bottom right graph-

http://forums.randi.org/showthread.php?t=76595&page=6


Nuts, I don't seem to be able to cut and paste the graphs here.


Note that there is a gray 'shaded' area on the graph. That is the range of all the proxies used. Not really shading but wiggly lines of a graph. See how it is such a uniform width an the left part of the graph, for about 600 years? In the midst of that range, we have the 'average', the dark line. Notice how from the years 1600 to 1900, the shading tapers down, until it they all merge into the dark line, and the proxies disappear. Then, for the last 100 years, we have only the 'average' line.

So, I have to wonder what happened to make all the proxies come into such a narrow range? Is there some fallacy there? Do proxies widen their range as they age? Do the rings in dead trees change as they age past 400 years? Changes to sedimentary layers? Does the co2 content of old ice change at the 300 year mark?

Is this "proxy merging" discussed elsewhere on the net?

My strength is NOT in the study of probability. But I do know that coins tossed 100 times have clusters of heads, and clusters of tails. If you actually flip a coin 100 times, the clusters are of about 8-9 in a row. That concept would show in this case as the width of the shaded area that represents the gross range of the proxies. Having all of the proxies come into agreement makes me think that the coin toss was rigged to come to one heads, one tails, one heads, one tail....

See, if the range of the proxies were to remain at it's former width, one would see simply that todays temperature is not so extreme, just one more high peak in another proxy. A peak that seems to have leveled off in the past few years.

[Number Six]

I don't know the answer to your question and I know nothing about proxy merging but when I see graphs like that the first thing I wonder is how accurately can they measure the termperature 100, 300, 500 years ago? I'm not sure how they do it, whether it's from tree rings or whatever, but it's one thing is being able to estimate it and another thing to know how much error there is in the measurement process. Maybe there do know how much error is in the measurement process and maybe it's a small error but what strikes me is I never hear it talked about, instead I just see graphs like that one that imply they know exactly what the earth temperature was way back when.

[jimbob]

My interpretation would be that since 1659, there have been some measurements of actual temperature, so one need rely less on the proxies...

http://www.metoffice.gov.uk/research/hadleycentre/obsdata/cet.html

As more measurements were taken in more locations, fewer proxies would be needed.

[casebro]

Yes but there are still fudge factors involved with the 'thermometer readings", they are not just direct readings. Some of the fudge factors are allowances for Urban Heat Islands, and even altitude of rural stations. So to my way of thinking, even the latest temp readings are just one more proxy, and all of the proxies ought to be shown on the graphs.

[dakotajudo]

Yes but there are still fudge factors involved with the 'thermometer readings", they are not just direct readings. Some of the fudge factors are allowances for Urban Heat Islands, and even altitude of rural stations. So to my way of thinking, even the latest temp readings are just one more proxy, and all of the proxies ought to be shown on the graphs.

I'm guessing you're confusing different kinds of measurement error.

Thermometer readings *are* direct readings; at least, as direct as we can make them. They are direct readings of air temperature, at a specific time and place.

Proxies on the other hand are direct measures of the outcome of processes that depend, at least in part, on the value to be determined.

Let's take, for example, a side track to bioassays; something I have a little more experience with. At one point, I wanted to directly measure the amout of cytokinin, a plant hormone, in tissue. Ideally, I would use a chemical assay - mix some chemicals and a color change indicates the presence of the compound of interest.

When you do this kind of assay, you, at the same time, set up a standard curve - mix known amounts of compound with your chemical, so you have a standard scale. Plot known values against colorimetric change, fit to a straight line using linear regression, then use that straight line (and associated formula) to predict compound concentration from color change.

Make sense?

With this line, there is a regression coefficient (r^2) - a measure of how well the line actually goes through the standard scale. Most chemical assays, you get a regression coefficient very close to 1 - 1 is good. - because most chemical assays involved very simple processes.

Sometimes, the chemical you are interested in is vary similar to other compounds (in the case of cytokinin, it's a lot like ATP). In this case, chemical assays don't work, because there's a lot of compounds will produce a similar chemical change.

So you use a bioassay - you add your unknown sample to some living system that responds to the compound of interest. In the cytokinin case there is a bioassay, using an Amaranth species commonly called red-root pigweed, that responds to cytokinin by producing more or less of a red pigment.

Measure the red pigment, you have a measure of the amount of cytokinin.

But not a vary accurate measure, since different plant samples won't respond to cytokinin exactly the same, every time.

So when you plot your standard curve - plant response vs. cytokinin level, you will see more deviation from a straight line, you will have a lower regression (or correlation coefficient). You will have much less accuracy in the value measured.

Still with me?


Well, now consider the case of estimating temperature based on tree ring width. Get an estimate of average yearly temperature (which you will probably get using thermometers?), measure tree rings widths for that year, then plot temperature vs width.

You get an estimate of temperature, but not something as accurate as direct temperature measurement. Tree rings are sensitive to temperature, but not to temperature only - water, for example, is pretty important in plant growth.

(Note that I'm glossing over the problems of sampling error for this - what part of the ring to you take for width, it's not even over the entire diameter - or calibration error - going from thermometer readings to tree rings back to temperature).

So, more simply, tree rings are a bioassay for temperature, and inherently not as accurate as direct thermometer readings.

As for carbon dioxide - do we know that carbon dioxide stays constant in ice over time - does it diffuse not at all, or only very slowly? If so, how slowly? If that relationship is not known, measurements further back in time become less accurate.

[Schneibster]

So nobody wants to discuss the "merging proxies"?Sure. They're all going up for the last several decades at a minimum. What precisely did you have in mind here? I'm having trouble seeing how all the proxies going up at the same time equates to buttheresnoglobalwarming. Maybe it's just me.

[CapelDodger]

So nobody wants to discuss the "merging proxies"?

So you're still trying to take refuge in the past?

The proxies are calibrated against actual observations over the last couple of centuries so it's hardly surprising that there's a "merging of proxies". The proxies themselves can be directly measured today instead of being extracted from ice-cores and such, with the inevitable loss of accuracy and detailed context.

Most proxies have only been calibrated against observations since the advent of cheap mass-spectroscopy, in the mid-70's. They have physical theory behind them as well, but that only gets you so far. Today we can directly measure, for instance, the isotopic oxygen content of fresh snow falling on a glacier. We can also observe the detailed context. We can't do that for the past, before direct measurements and global surveillance were available.

On another thread (damned if I can keep track of them all) you chose to represent historic temperatures by selecting single and wildly excursionary proxies as truly representative of the global temperature of the time. All of which, unsurprisingly, were selected because they indicated a temperature greater than today's back before the Little Ice Age. Time was that the LIA was the battleground where proxies were chosen for their coldness, but things didn't go so well there, so now salvation is sought in the Medieval Period. At that exponential rate you'll be seeking refuge in the Permian before you're middle-aged.

[mhaze]

Yes but there are still fudge factors involved with the 'thermometer readings", they are not just direct readings. Some of the fudge factors are allowances for Urban Heat Islands, and even altitude of rural stations. So to my way of thinking, even the latest temp readings are just one more proxy, and all of the proxies ought to be shown on the graphs.

By "merging proxies" you are referring to the 70 or so datasets from different sources - tree rings, ice core, etc. that were all averaged, or by some other statistical procedure, combined together to form the basis for the historical record in some of the graphs we see?

Is that correct?

[CapelDodger]

Sure. They're all going up for the last several decades at a minimum. What precisely did you have in mind here? I'm having trouble seeing how all the proxies going up at the same time equates to buttheresnoglobalwarming. Maybe it's just me.

I think what casebro has in mind here, on yet another AGW thread he's spawned, is revealed in the first word of the OP : "So". casebro has reached yet another cast-iron interpretation which he'll abandon as soon as he gets confused about it. Which is pretty damn soon.

So nobody wants to discuss the "merging proxies"? Not from the evidence of my own lying eyes. A thread concerning the Swindlers is obviously not the venue. This one is.

[casebro]

Can somebody link me to a graph of a bunch of proxies up to say, 2000?

[Beausoleil]

You mean "proxies for temperature", presumably?

In order to convert a "proxy for temperature" into temperature, you need a range over which readings of both are available. From this you calculate a function that turns converts the proxy into a temperature. The function will contribute an error to the derived temperature reading.

I suspect that the modern dataset is the range used for defining how to convert a proxy for temperature into actual temperature. So the graph you are looking for will show them all agreeing by definition. In other words, it's pointless.

(That's my guess, anyway.)

[mhaze]

This chart and the accompanying notes are from www.globalwarmingart.com (http://www.globalwarmingart.com).

The notes say that the data is from (1) Jones and refs therein, this makes it somewhat suspicious IMHO. But this chart does show the phenomena of converging proxy data you mentioned, so it's a start.



http://forums.randi.org/imagehosting/thum_14224469a1ba13420f.png (http://forums.randi.org/vbimghost.php?do=displayimg&imgid=7039)


This image is a comparison of 10 different published reconstructions of mean temperature changes during the last 1000 years. More recent reconstructions are plotted towards the front and in redder colors, older reconstructions appear towards the back and in bluer colors. An instrumental history of temperature is also shown in black. The medieval warm period (http://en.wikipedia.org/wiki/medieval_warm_period) and little ice age (http://en.wikipedia.org/wiki/little_ice_age) are labeled at roughly the times when they are historically believed to occur, though it is still disputed whether these were truly global or only regional events. The single, unsmoothed annual value for 2004 is also shown for comparison. (Image:Instrumental Temperature Record.png (http://www.globalwarmingart.com/wiki/Image:Instrumental_Temperature_Record_png) shows how 2004 relates to other recent years).
It is unknown which, if any, of these reconstructions is an accurate representation of climate history; however, these curves are a fair representation of the range of results appearing in the published scientific literature. Hence, it is likely that such reconstructions, accurate or not, will play a significant role in the ongoing discussions of global climate change (http://en.wikipedia.org/wiki/climate_change) and global warming (http://en.wikipedia.org/wiki/global_warming).
For each reconstruction, the raw data has been decadally smoothed with a σ = 5 yr Gaussian weighted moving average. Also, each reconstruction was adjusted so that its mean matched the mean of the instrumental record during the period of overlap. The variance (i.e. the scale of fluctuations) was not adjusted (except in one case noted below).
Except as noted below, all original data for this comparison comes from [1] (http://www.ngdc.noaa.gov/paleo/recons.html) and links therein. It should also be noted that many reconstructions of past climate report substantial error bars, which are not represented on this figure.





(dark blue 1000-1991): [abstract (http://www.globalwarmingart.com/wiki/Reference:Jones_et_al._1998)] [DOI (http://dx.doi.org/10.1191%2F095968398667194956)] Jones, P.D., K.R. Briffa, T.P. Barnett, and S.F.B. Tett (1998). "High-resolution Palaeoclimatic Records for the last Millennium: Interpretation, Integration and Comparison with General Circulation Model Control-run Temperatures". The Holocene 8: 455-471.
(blue 1000-1980): [abstract (http://www.globalwarmingart.com/wiki/Reference:Mann_et_al._1999)] [full text (http://www.ncdc.noaa.gov/paleo/pubs/millennium-camera.pdf)] Mann, M.E., R.S. Bradley, and M.K. Hughes (1999). "Northern Hemisphere Temperatures During the Past Millennium: Inferences, Uncertainties, and Limitations". Geophysical Research Letters 26 (6): 759-762.
(light blue 1000-1965): [abstract (http://www.globalwarmingart.com/wiki/Reference:Crowley_and_Lowery_2000)] Crowley, Thomas J. and Thomas S. Lowery (2000). "Northern Hemisphere Temperature Reconstruction". Ambio 29: 51-54. ; Modified as published in [abstract (http://www.globalwarmingart.com/wiki/Reference:Crowley_2000)] [DOI (http://dx.doi.org/10.1126%2Fscience.289.5477.270)] Crowley (2000). "Causes of Climate Change Over the Past 1000 Years". Science (http://en.wikipedia.org/wiki/Science_%28journal%29) 289: 270-277.
(lightest blue 1402-1960): [abstract (http://www.globalwarmingart.com/wiki/Reference:Briffa_et_al._2001)] [DOI (http://dx.doi.org/10.1029%2F2000JD900617)] Briffa, K.R., T.J. Osborn, F.H. Schweingruber, I.C. Harris, P.D. Jones, S.G. Shiyatov, and E.A. Vaganov (2001). "Low-frequency temperature variations from a northern tree-ring density network". J. Geophys. Res. 106: 2929-2941.
(light green 831-1992): [abstract (http://www.globalwarmingart.com/wiki/Reference:Esper_et_al._2002)] [DOI (http://dx.doi.org/10.1126%2Fscience.1066208)] Esper, J., E.R. Cook, and F.H. Schweingruber (2002). "Low-Frequency Signals in Long Tree-Ring Chronologies for Reconstructing Past Temperature Variability". Science (http://en.wikipedia.org/wiki/Science_%28journal%29) 295 (5563): 2250-2253.
(yellow 200-1980): [abstract (http://www.globalwarmingart.com/wiki/Reference:Mann_and_Jones_2003)] [full text (http://stephenschneider.stanford.edu/Publications/PDF_Papers/mannjones03-preprint.pdf)] [DOI (http://dx.doi.org/10.1029%2F2003GL017814)] Mann, M.E. and P.D. Jones (2003). "Global Surface Temperatures over the Past Two Millennia". Geophysical Research Letters 30 (15): 1820.
(orange 200-1995): [abstract (http://www.globalwarmingart.com/wiki/Reference:Jones_and_Mann_2004)] [full text (http://holocene.meteo.psu.edu/shared/articles/JonesMannROG04.pdf)] [DOI (http://dx.doi.org/10.1029%2F2003RG000143)] Jones, P.D. and M.E. Mann (2004). "Climate Over Past Millennia". Reviews of Geophysics 42: RG2002.
(red-orange 1500-1980): [abstract (http://www.globalwarmingart.com/wiki/Reference:Huang_2004)] [DOI (http://dx.doi.org/10.1029%2F2004GL019781)] Huang, S. (2004). "Merging Information from Different Resources for New Insights into Climate Change in the Past and Future". Geophys. Res Lett. 31: L13205.
(red 1-1979): [abstract (http://www.globalwarmingart.com/wiki/Reference:Moberg_et_al._2005)] [full text (http://stephenschneider.stanford.edu/Publications/PDF_Papers/MobergEtAl2005.pdf)] [DOI (http://dx.doi.org/10.1038%2Fnature03265)] Moberg, A., D.M. Sonechkin, K. Holmgren, N.M. Datsenko and W. Karlén (2005). "Highly variable Northern Hemisphere temperatures reconstructed from low- and high-resolution proxy data". Nature 443: 613-617.
(dark red 1600-1990): [abstract (http://www.globalwarmingart.com/wiki/Reference:Oerlemans_2005)] [DOI (http://dx.doi.org/10.1126%2Fscience.1107046)] Oerlemans, J.H. (2005). "Extracting a Climate Signal from 169 Glacier Records". Science 308: 675-677.

[casebro]

Interesting graph. I just scanned each abstract, and every line but #10 (red) is an 'average' of many inputs already. #10 is strictly glaciers. 4&5 are tree rings, but of many series. Other lines used wider data sources. Each line has had many spikes averaged out. So, if we consider the 'thermometer series' on the same basis, then we have not yet averaged the recent spike.

Likewise, each line used several inputs. How many lines used the same inputs? Wouldn't using common inputs result in a common end result?

Does it seem odd that with all the historical variations in the data, that NONE of the proxies chosen as examples show any variation from the thermometer series?

Or, lets look at this graph from right to left, recent to ancient. All the proxies start out as agreeing with the current temperature series. The graph shows that ALL of the proxies get more erratic as they try go back in time. If so, then we really don't have any dependable basis to compare the current temps to.

[mhaze]

I've purposely avoided all the arguments on little ice age, medieval warming period, holocene and the like - all the interpretations of proxy data, because it truly seems like a "he said she said" argument. Also it looked bogus.

Take for example another thread on possible fraud in the Jones data series. Reading Jones, I find that he thought that for Australia, taking the minimum and maximum daily temperature and averaging them would give him a satisfactory "average temperature". He then proceeds to use that "average" in his work for urban heat islands.

I see a real problem with (T max + T min)/2 = T average and then combining that data series with others that have a true daily average, and believe that anyone who would actually publish such work does not know what they are doing.

You are asking about a more complicated statistical issue with longer range trends. On this forum flame baiting and trolls may make that discussion impossible.

[Schneibster]

Confusion achieved. Thermometers now no longer work. Nice call, CD.

[mhaze]

Interesting graph. I just scanned each abstract, and every line but #10 (red) is an 'average' of many inputs already. #10 is strictly glaciers. 4&5 are tree rings, but of many series. Other lines used wider data sources. Each line has had many spikes averaged out. So, if we consider the 'thermometer series' on the same basis, then we have not yet averaged the recent spike.

Likewise, each line used several inputs. How many lines used the same inputs? Wouldn't using common inputs result in a common end result?

Does it seem odd that with all the historical variations in the data, that NONE of the proxies chosen as examples show any variation from the thermometer series?

Or, lets look at this graph from right to left, recent to ancient. All the proxies start out as agreeing with the current temperature series. The graph shows that ALL of the proxies get more erratic as they try go back in time. If so, then we really don't have any dependable basis to compare the current temps to.

And the error bars were left out.

[a_unique_person]

Confusion achieved. Thermometers now no longer work. Nice call, CD.

Do we really know anything at all? The triumph of sophistry.

[mhaze]

Interesting graph. I just scanned each abstract, and every line but #10 (red) is an 'average' of many inputs already. #10 is strictly glaciers. 4&5 are tree rings, but of many series. Other lines used wider data sources. Each line has had many spikes averaged out. So, if we consider the 'thermometer series' on the same basis, then we have not yet averaged the recent spike.

Likewise, each line used several inputs. How many lines used the same inputs? Wouldn't using common inputs result in a common end result?

Does it seem odd that with all the historical variations in the data, that NONE of the proxies chosen as examples show any variation from the thermometer series?

Or, lets look at this graph from right to left, recent to ancient. All the proxies start out as agreeing with the current temperature series. The graph shows that ALL of the proxies get more erratic as they try go back in time. If so, then we really don't have any dependable basis to compare the current temps to.

The easy way to answer your questions is really to email one or several of the scientists, alternately, to look at discussions where this subject has been addressed on other websites.

http://www.climateaudit.org/?page_id=354

Pretty good list of merged/multiproxy studies in pdf form, all downloadable.

[mhaze]

Does it seem odd that with all the historical variations in the data, that NONE of the proxies chosen as examples show any variation from the thermometer series?

Or, lets look at this graph from right to left, recent to ancient. All the proxies start out as agreeing with the current temperature series. The graph shows that ALL of the proxies get more erratic as they try go back in time. If so, then we really don't have any dependable basis to compare the current temps to.

Surprise.

Did you know all the proxy series (proxies are things like tree ring growth from which temperature is inferred) stop at 1980? From then on it's our friendly thermometers. Climateaudit makes an interesting point about this - if we want to believe in the valid historical readings of the proxies, shouldn't they be brought up to date?

Proxy readings for 1998 should be off the chart, then, since 1998 was the hottest year on record. Right?

The discussion at climateudit.org is here, (http://www.climateaudit.org/?p=89#more-89) and it appears to have originated on the realclimate website. Briefly,

One of the first question that occurs to any civilian becoming familiar with these studies (and it was one of my first questions) is: what happens to the proxies after 1980? Given the presumed warmth of the 1990s, and especially 1998 (the warmest year in the millennium�), you'd think that the proxy values would be off the chart. In effect, the last 25 years have provided an ideal opportunity to validate the usefulness of proxies and, especially the opportunity to test the confidence intervals of these studies, put forward with such assurance by the multiproxy proponents. What happens to the proxies used in MBH99 or Moberg et al [2005] or Crowley and Lowery [2000] in the 1990s and, especially, 1998?


This question about proxies after 1980 was posed by a civilian to Mann in December at realclimate (http://www.realclimate.org/index.php?p=11#comments). Mann replied:Most reconstructions only extend through about 1980 because the vast majority of tree-ring, coral, and ice core records currently available in the public domain do not extend into the most recent decades. While paleoclimatologists are attempting to update many important proxy records to the present, this is a costly, and labor-intensive activity, often requiring expensive field campaigns that involve traveling with heavy equipment to difficult-to-reach locations (such as high-elevation or remote polar sites). For historical reasons, many of the important records were obtained in the 1970s and 1980s and have yet to be updated. [my bold]

Trillions of dollars of projected tax increases, $46B spent recently for climate research in the US, $360B on Kyoto in Europe, 3.4B spent in Australia to reduce temperature by a whopping 0.000025 C...and we're getting told this is a costly and labor-intensive activity?

You'd think everyone would be working together to figure out the facts.

Surreal.....

[casebro]

Yeah, how much does it cost to drill a hole in a tree? It wouldn't even have to be a very deep hole. Times 200?

"Converging proxies" is probably a better term. Less likely to be confused with the fact that each of the proxies are made by 'merging' several 'series' of data. It just seems strange to me that the range of the proxies changes from .2 in 1950 to .7 degree C in 1650 . Notice also that there are 10 proxies in 1650, but only four in the year 1000. I'd guess that if all 10 were extended back, the variation in 1000 would be what- close to THREE degrees?

Well, I guess you would have to calibrate your proxies to known temps. So during the calibration period, they all should be similar. But then, as we project the proxies back into time, we see that they show such dis-similarity that how can we trust them? Don't we really know the average temp of the year 900 to within a range of 3 degrees? So how important is the .6 degree rise, from the thermometer series?

If each series has an acuity of one degree, and you have five series, the overall acuity can NOT get better than one degree. It can only be as tight as one degree if all series converge perfectly. And the extreme range is only limited by the total variation between series, plus that one degree.

So, what is the acuity of any series?

So it looks like they took over two hundred series of data, and merged them into proxies. Then they took all of those proxies, and averaged them together into one line, the 'trend' line. It looks really technical, but the reality is that is no way to refine any actual data points. The data is still only accurate to a range of three degrees, and I suppose it only gets worse as we try to project back thousands of years.

So how can we say that it is warmer today than ever before? Or that the rate of warming is faster?

[mhaze]

When you first posted the question I understood the problem because it is similar to some others I have had to confront. However, I didn't know the answer in a qualitative sense. It is necessary to go step by step through the problem and know the error bounds at each step.

Is the point of this that you would like to assert that the old data is not conclusive, or to prove it? These are different things of course.

It is not a good thing when an amateur can easily find errors in published, peer reveiwed articles....but ain't the internet, that gives us this capability, downright fun?

[Beausoleil]

Or, lets look at this graph from right to left, recent to ancient. All the proxies start out as agreeing with the current temperature series.



Of course. To convert a proxy into a temperature you have to use a sequence over which both have been measured to deduce a calibration function. When you apply that function to the proxy values in the calibration series, of course you reproduce the temperature values in the calibration series.



The graph shows that ALL of the proxies get more erratic as they try go back in time. If so, then we really don't have any dependable basis to compare the current temps to.


Non sequitur.

[casebro]

Of course. To convert a proxy into a temperature you have to use a sequence over which both have been measured to deduce a calibration function. When you apply that function to the proxy values in the calibration series, of course you reproduce the temperature values in the calibration series.

I think I covered that in my latest post, #20, para 3.


Non sequitur.

If you say so.

Using the graph in post 12, You tell me just how warm it was in 1650. With a tolerance figure.

[mhaze]

http://web.utk.edu/~grissino/default.html

Tree Ring Proxy site with software descriptions

[casebro]

Lots of linkages on that site. But I could not find any mention of the acuity/tolerance of retro-casted* data. I did find that Mann's first graph showed a very broad area, probably the 90% probability 'zone', which showed historical spikes as tall as the current thermometer series shows.

Generally, tree rings show CLIMATE, not temperature. It takes quite a leap to convert the rings to temperature- with what acuity?

*whats a better word? the tree ring data can be matched to the temp series for the time they overlap, going backwards in time from there is a 'backwards forecast', or a _________?

[mhaze]

Lots of linkages on that site. But I could not find any mention of the acuity/tolerance of retro-casted* data. I did find that Mann's first graph showed a very broad area, probably the 90% probability 'zone', which showed historical spikes as tall as the current thermometer series shows.

Generally, tree rings show CLIMATE, not temperature. It takes quite a leap to convert the rings to temperature- with what acuity?

*whats a better word? the tree ring data can be matched to the temp series for the time they overlap, going backwards in time from there is a 'backwards forecast', or a _________?



I posed the following question on climateaudit.org:
My question is on tree rings: I would like to know if there are what may be called “error bounds” on temperatures derived from tree rings, and if so, how they are calculated. I’ve found numerous explanations of determining temp from rings, but nothing that said “So the temp for 1703 is known to be 19.6 C to 19.95 C and here is why”. Also, do tree ring temp estimates become less certain with older series, or are they considered to be relatively consistant as to the error bounds?



There were two responses. Here is the first (bold is mine) -



See http://www.climateaudit.org/?p=647 ,how they are calculated
in the MBH99 case nobody knows..
I tried to compute error bounds for some known proxy sets using Brown’s Multivariate Calibration as a reference:
http://signals.auditblogs.com/2007/07/09/multivariate-calibration-ii/
These results indicate that proxy data is not accurate enough to make conclusions about unprecedentedness of 20th century. And maybe here’s an explanation why there’s no interaction between the Team and mainstream statisticians. But this is all non-peer reviewed, so let’s not jump to conclusions too early.



A second response was more to the point -


Cores from trees are like cores from ice. They tell you something, but I have no idea what that something is

[rockoon]

*whats a better word? the tree ring data can be matched to the temp series for the time they overlap, going backwards in time from there is a 'backwards forecast', or a _________?

Not precisely the same, but when using a model (for example) and wish to "verify" its "skill", you would use "hindcasting" or "aftcasting" .. that is to say, have it try to predict known events of the past.

The danger of this methodology is that there are an infinite number of models that will show "skill" at predicting any arbitrary sequence, with only a small set of those having actual skill rather than coincedental.

In the strict sense I do not believe there is a word for the process you describe that includes "cast" because quite frankly, it isnt any form of prediction in the normal sense.

I would question on what grounds each proxy are weighted, if they are at all. They certainly should be weighted based on their skill at reflecting local temperatures. Also, the proxies should only be used in terms of what they are, proxy measurements of local temperatures. That means that you cannot use the bristlecone proxies at if they were proxies for hemispheric temperatures (Sorry Mann, you don't get to do that, so stop doing it!)

[casebro]

The whole point is that NONE of the 'hindcasts' are so accurate that anybody can claim that either:

a) it is hotter now than________.
b) it is getting hotter FASTER than ever.

The "Thermometer Series" that shows our current one degree rise, is well within the parameters of all those other series.

[mhaze]

The whole point is that NONE of the 'hindcasts' are so accurate that anybody can claim that either:

a) it is hotter now than________.
b) it is getting hotter FASTER than ever.

The "Thermometer Series" that shows our current one degree rise, is well within the parameters of all those other series.

One point of view.

A scientific article purporting to show past reconstructions should itself explain and substantiate its methods, provide algorithms and data to allow independent verification.

Or its junk science?

[rockoon]

A scientific article purporting to show past reconstructions should itself explain and substantiate its methods, provide algorithms and data to allow independent verification.

Should they explain their methods? Yes.
Should they provide data? Absolutely.
Should they provide algorithms? Isnt this the same as explaining their methods?
Should they substantiate their methods? Isn't that what independent verification achieves?


Or its junk science?

..or simply "elite" science. Wanting to get credit for it while at the same time keeping things, for whatever reason, a secret.

[varwoche]

I posed the following question on climateaudit.org ... See http://www.climateaudit.org/?p=647 The exchange is not on the page you linked to.

(I'm just about at my wild goose threshold (http://forums.randi.org/showthread.php?postid=2835664#post2835664) with your links.)

[mhaze]

Should they explain their methods? Yes.
Should they provide data? Absolutely.
Should they provide algorithms? Isnt this the same as explaining their methods?
Should they substantiate their methods? Isn't that what independent verification achieves?



..or simply "elite" science. Wanting to get credit for it while at the same time keeping things, for whatever reason, a secret.


Here is an excerpt from the classic McIntyre and McKitrick (2003) critique of Mann et. all. 1998 (the "Hockey Stick" paper).

Mann is a multiproxy study which is very representative of the problem discussed in this thread. But in going back and reading Mann 1998, and MM2003, not only does one wind up with question marks all over the place regarding the issue of precision of the backwards forcasts, but more basic issues. Example from the critique by McIntyre follows.


http://forums.randi.org/imagehosting/thum_1422446b86f272835d.png (http://forums.randi.org/vbimghost.php?do=displayimg&imgid=7492)

The first series is from Mann 1998. The third is the actual data series untruncated. A reaonable interpretation:

A. Mann wants the current warming to be the worst in history, resulting in "unprecendented man made global warming".
B. He or his staff truncate data points from year 1525 to 1550.
C. Result: current warming worst in history.

References.
MM2003 (http://www.climateaudit.org/pdf/mcintyre.mckitrick.2003.pdf)
MBH1998

MBH1998 link shows the "server busy" but the paper should be here. (http://forums.randi.org/holocene.evsc.virginia.edu)
holocene.evsc.virginia.edu

[David Rodale]

This 2005 paper discusses proxies and tree ring data:
http://www.ifir.edu.ar/~redes/ps/EPSL2005.pdf

[casebro]

Anybody know exactly what todays 'global average temperature' is, in degrees? Or any particular point in recent time?

[mhaze]

Anybody know exactly what todays 'global average temperature' is, in degrees? Or any particular point in recent time?

how about combining lighthouse and ASOS data?

lighthouse data goes way back, few adjustments, no UHI
Asos is modern hourly automated airport observations.

[casebro]

No takers on a current temperature? Only that "it's hotter than ever"?

[Geckko]

This is probably a salutory lesson in how we use statistics in the climate change debate:

"1998 no longer the hottest year on record in USA" (http://www.norcalblogs.com/watts/2007/08/1998_no_longer_the_hottest_yea.html)

[rockoon]

I believe McIntyre is one of the most honest men in climate science, and its not even his field.

<insert defamation from a warmer that doesnt know the difference between McKitrick and McIntyre below this post>

[mhaze]

Top 10 GISS U.S. Temperature deviation (deg C) in New Order 8/7/2007
Year Old New 1934 1.23 1.25 1998 1.24 1.23 1921 1.12 1.15 2006 1.23 1.13 1931 1.08 1.08 1999 0.94 0.93 1953 0.91 0.90 1990 0.88 0.87 1938 0.85 0.86 1939 0.84 0.85
I knew from looking the members up a month buck that the 19305 were way hotter than recent temps. But this is not over yet-aren't there 1000 of the Stations left to Check?

Headline?

Blogger's Boots on the Ground Trumps NASA's Eyes In the Sky!

[mhaze]

Tablet PC hopelessly scrambles words again.

I knew from looking the members up a month buck that the 19305 were way hotter than recent temps. But this is not over yet-aren't there 1000 of the Stations left to Check?:confused:

Should be

I knew from looking the numbers up a month ago that the 1930s were way hotter than recent temps. But this is not over yet-aren't there 1000 of the stations left to check?

[mhaze]

I believe McIntyre is one of the most honest men in climate science, and its not even his field.



Looks like he's quite popular this morning.

www.climateaudit.org overloaded with traffic!

[mhaze]

Looks like he's quite popular this morning.

www.climateaudit.org (http://www.climateaudit.org) overloaded with traffic!

Something of a mirror at icecap, (http://icecap.us/) if anyone wants to read about this.

[JoeEllison]

I believe McIntyre is one of the most honest men in climate science, and its not even his field.

<insert defamation from a warmer that doesnt know the difference between McKitrick and McIntyre below this post>
There's one born every minute... which minute were you?:p

[BobK]

Mhaze,
Evidently someone doesn't like having the temperature become cooler.

ClimateAudit is experiencing a DDOS attack today. I received an email today about it.

Dear CA reader

CA has been knocked off the internet by a DDOS attack. We are going to move the CA domain to a temporary page while I move the CA files and databases to a new server behind a much better firewall.

Its obvious that someone can't take constructive criticism.

We should be back in a few days. snip ...

John
McIntyre is the one that caused GISS to make wholesale changes to temperatures over the past couple of days. He pointed out a flaw that required GISS to lower temperatures during this current century by up to .15 degrees, and now I guess they have to adjust many of last century's values also. Things probably won't shake out for a while.

[mhaze]

Anthony Watts, www.surfacestations.org is also off line.

[rockoon]

There's one born every minute... which minute were you?:p

The minute isnt as important as the day because I wasn't born yesterday.

:D

[David Rodale]

Anthony Watts, www.surfacestations.org is also off line.



According to Anthony, his had DoS attacks a few weeks ago, but today he's not sure.

Climate Audit is definitely a monstrous DoS.

[mhaze]

http://science.slashdot.org/article.pl?sid=07/08/10/1530251&from=rss

Steve. (http://science.slashdot.org/article.pl?sid=07/08/10/1530251&from=rss) Slashdotted (http://science.slashdot.org/article.pl?sid=07/08/10/1530251&from=rss)

[rockoon]

You would think that a shill for the oil companies that purportedly spend millions on disinformation would have a site that could handle being slashdotted.

(maybe this is all part of their ruse.. to appear indpendent you must have a low budget site)

[mhaze]

Reynolds at Instapundit seems to be taking the NASA bug-catch quite seriously

This -
ACE WONDERS WHY NOBODY'S TALKING ABOUT the NASA climate data revision. (http://minx.cc/?post=236691)
UPDATE: Well, here's a bit of notice. (http://www.opinionjournal.com/best/?id=110010455#y2k)
ANOTHER UPDATE: More here: (http://www.ecotality.com/blog/2007/oops/) "Will the mainstream media report the corrected story with as much gusto as they initially reported the claim that 1998 was the warmest on record? Doubtful. But they should. Good public policy can not be made on bad data."
MORE: This (http://www.ecotality.com/blog/2007/oops/#comment-21595) comment at Ecotality distinguishes hottest years in America from hottest years globally, but I always understood this to be about American, not global, records. And I think I was right. As I noted in my earlier post, (http://instapundit.com/archives2/008011.php) it indicates problems with the data sets. More here: (http://www.coyoteblog.com/coyote_blog/2007/08/official-us-cli.html) The GISS today makes it clear that these adjustments only affect US data and do not change any of their conclusions about worldwide data. But consider this: For all of its faults, the US has the most robust historical climate network in the world. If we have these problems, what would we find in the data from, say, China? And the US and parts of Europe are the only major parts of the world that actually have 100 years of data at rural locations. No one was measuring temperature reliably in rural China or Paraguay or the Congo in 1900. That means much of the world is relying on urban temperature measurement points that have substantial biases from urban heat.
Much more information at the link.
Plus, reports of Denial-of-Service attacks. (http://www.classicalvalues.com/archives/2007/08/climate_audit_h.html)


Also from Reynolds
(http://www.classicalvalues.com/archives/2007/08/climate_audit_h.html)
KIND OF A COUP: The Amazon page for Bjorn Lomborg's new book on global warming (http://www.amazon.com/gp/product/0307266923?ie=UTF8&tag=wwwviolentkicom&link_code=as3&camp=211189&creative=373489&creativeASIN=0307266923) features a review by Michael Crichton.
posted at 10:59 AM by Glenn Reynolds http://www.instapundit.com/graphics/permalink.gif (http://instapundit.com/archives2/008072.php)