How do we know the temperature thousands of years ago?

[blike]

A friend of mine brought up the objection to global warming that we simply have more accurate thermometers now. That got me thinking, how can we eliminate that possibility? How do we know the genereal temperature trends prior to the invention of thermometers?

[insane_alien]

something to do with the ratios of oxygen isotopes trapped in ice i believe.

not quite sure why this ratio has to do with temperature but i'm pretty sure they must have a good idea of what they are doing.

[ecoli]

http://www.worldviewofglobalwarming.org/pages/paleoclimate.htm

[swansont]

ecoli's link mentions the oxygen but not the details. You have O-16 and O-18 (and a little O-17), and it takes more energy to vaporize O-18 because it's more massive. It also condenses more easily. So there is a correlation with temperature, as more O-18 shows up in sediments and ice layers when it's colder.

[SkepticLance]

I would word the problem a little differently. There are lots of ways of reading fossil records of temperature. I would be more concerned about the comparability of those results with modern thermometers. Changing measuring technique in the middle of a study can distort the results.

 

If you can easily download the following reference (I have an old fashioned slow system), you will see two graphs. Above graph changes measurement method mid study. Bottom graph carries the same method right through (tree rings).

 

http://putfile.com/pic.php?pic=6/16504473590.jpg&s=f5

 

You may care to comment on how the two sets of results differ.

[timo]

If you can easily download the following reference (I have an old fashioned slow system)' date=' you will see two graphs. Above graph changes measurement method mid study. Bottom graph carries the same method right through (tree rings).

 

http://putfile.com/pic.php?pic=6/16504473590.jpg&s=f5

 

You may care to comment on how the two sets of results differ.[/quote']

What´s the calculation rule to convert "mm tree-ring growth" to "relative temperature (wrt to some average, I guess) in degrees of celsius"? Is comparison of the two curves possible without knowing how they are related?

[SkepticLance]

Atheist

I am not an expert on dendrochronology, so cannot answer your first question.

 

Your second question is the crux. I do not believe you can directly compare studies using two different measuring techniques. The top graph uses tree rings (as does the bottom graph) until about the year 1890, then switches to direct thermometer measurements. I do not believe that the fact the graph suddenly changes at that point is coincidence.

[swansont]

I would word the problem a little differently. There are lots of ways of reading fossil records of temperature. I would be more concerned about the comparability of those results with modern thermometers. Changing measuring technique in the middle of a study can distort the results.

 

If you can easily download the following reference (I have an old fashioned slow system)' date=' you will see two graphs. Above graph changes measurement method mid study. Bottom graph carries the same method right through (tree rings).

 

http://putfile.com/pic.php?pic=6/16504473590.jpg&s=f5

 

You may care to comment on how the two sets of results differ.[/quote']

 

 

Why are they different pre-1890, if they used the same method to that point?

[DrCloud]

The general answer to the original question is "proxy data." By this, it's meant that various indicators preserved over time are used to infer ancient temperatures -- oxygen isotope ratios and tree-ring widths being a couple. Others include distributions of critters and plant material in sediments and ice.

 

Necessarily, these inferred temperatures are both less precise and less accurate than modern thermometers; nonetheless, they are self-consistent and useful in reconstructing the behavior (particularly relative behavior) of past climates.

 

It's important to understand the difficulty of this: not only is it necessary to translate the proxies into temperatures (or, more often, temperature deviations from some base), but the timing of the temperatures must also be established. After all, thermometers aren't the only thing that's new in geological terms, so are clocks. So getting the time scale right for the inferred temperatures is also quite challenging and also leads to additional uncertainties in absolute terms.

 

Because of these uncertainties, some people enjoy minimizing the usefulness of proxy datasets. This, however, speaks more to their lack of knowledge about the field of paleoclimatology than it does to the datasets themselves. Understanding Earth's past behavior requires using all the information we have available, in the context of its limitations, and proxy data is all we have in the era before thermometers. HPH

[SkepticLance]

Swansont said :

 

Why are they different pre-1890, if they used the same method to that point?

 

Swansont. You are being deliberately difficult. You know enough about this subject so that 5 seconds thought would have given you the answer to this question.

 

For other people who may know less, I will answer your question.

 

There are literally dozens of studies that claim to measure temperature change over the last 500 to 1000 years. Each and every one of them comes up with different values, which makes the two studies I gave quite normal in their differences.

 

The reason for different values is simply a result of the nature of what they are studying. They try to derive global average temperature. To do that, they would need an unlimited number of samples from an unlimited number of sites to get a statistically perfect average. Of course they do not do that. Each study uses a 'small' number of samples from a limited number of sites. Local conditions then skew the results. For example : an ice core sample from Greenland in the year 1750 may give a different result to a tree ring sample in the year 1750 from Britain. The summer/winter temperatures from the two sites will not be the same. While each study attempts to get enough samples to overcome this source of error, no-one actually succeeds.

 

However, these are smaller difference - one study to another. The point I was making is that the difference between the two studies I presented was very dramatic after the year 1890. That is when they changed measurement technique in the Mann study, while the second study kept a consistent measurement method.

 

My suspicion is that the Mann study introduced a major source of error by changing technique. I suspect the error was actually pre-1890, with temperature changes as inferred being somewhat less than reality.

 

As evidence for this, I would point out the small difference as shown for the Medieval warm period, versus the Little Ice Age. Historical data shows that these were dramatic warmings and coolings. However, they are shown as minimal in the Tree Ring studies. This is always a possible problem in masurements that are indirect.

[Chuck Edge]

A friend of mine brought up the objection to global warming that we simply have more accurate thermometers now. That got me thinking, how can we eliminate that possibility? How do we know the genereal temperature trends prior to the invention of thermometers?

 

I imagine you start with an equation of state with temperature as a parameter and build a more complex model around it with things of interest to paleoclimatologists as inputs. But good question.

[DrCloud]

^ In the sense that "equation of state" refers to the behaviors of foraminifera, corals, flowering plants (pollen), and trees, yes. But not really in the sense of physics and chemistry. HPH

[swansont]

Swansont said :

 

Why are they different pre-1890' date=' if they used the same method to that point?[/i']

 

Swansont. You are being deliberately difficult. You know enough about this subject so that 5 seconds thought would have given you the answer to this question.

 

And yet you presented the two graphs for comparison. Why do that if they weren't measuring the same thing?

 

Different techniques yielding different answers is unreliable, somehow, but the same technique yielding different answers is not, because the measurements are localized. And yet, localized measurements could still be the reason for the differences in the graphs. You haven't presented any evidence that makes the case for one scenario over the other.

[SkepticLance]

Swansont.

Take another look at the two graphs. You and I have both seen numerous graphs of temperature change over the last 500 to 1000 years. We have seen that they are all different to each other, but in relatively minor ways, as are the two graphs I presented.

 

However, the differences after 1890 are not minor. This is a dramatic change.

This represents an extra factor in the equation.

[swansont]

Swansont.

Take another look at the two graphs. You and I have both seen numerous graphs of temperature change over the last 500 to 1000 years. We have seen that they are all different to each other' date=' but in relatively minor ways, as are the two graphs I presented.

 

However, the differences after 1890 are not minor. This is a dramatic change.

This represents an extra factor in the equation.[/quote']

 

 

Considering that there is no calibration given, i.e. the top graph is in degrees C and the bottom in mm of growth, I don't see how any one conclusion can be supported. The tree-ring graph shows cooling from 1000 to 1300, spanning the entire range of growth values, no less, while the other graph is flat. That's pretty dramatic, too. But no way of knowing what temperature range it represents, so there's no way to tell if it's a big as the upper curve's change. Absent a calibration it's apples and oranges.

[SkepticLance]

Swansont.

You quibble. You wriggle like a fish on a hook. All this is irrelevent.

 

My sole point is that changing the method of measurement changes the results. That point stands out like the Taj Mahal.

[swansont]

Swansont.

You quibble. You wriggle like a fish on a hook. All this is irrelevent.

 

My sole point is that changing the method of measurement changes the results. That point stands out like the Taj Mahal.

 

 

But you haven't actually demonstrated that. You need evidence. Show the calibration for the tree-growth curve, show that it's global data and not local, and then you're on to something. But until that or similar evidence is presented, you simply can't draw the conclusion that you have.

[SkepticLance]

Swansont.

Both graphs vary by no more than 0.2C from 1000 AD to 1890. The Mann graph then shows a surge of 0.7C.

 

Again, you are avoiding the main point. And that is that the Mann graph shows a major change in results 'coinciding' with a change in measurement technique, while the other graph, which keeps the same measurement method, shows no significant change at that point. I know you get the point, and are just raising quibbles to deny it to yourself.

 

I have thought about the significance of these results. I have to regard both graphs with scepticism, since the change in temperature shown from (say) 1200 to 1600 AD is minimal. Yet we know from historical data that there was a relatively massive drop in temperature over those years. In the 13th Century in Europe, there was widespread crop failure and deaths by starvation due to temperature drop. Neither graph shows this, which makes me think the tree ring method probably understates temperature change.

 

Thus my interpretation is that the results prior to 1890 for the Mann graph, and all results for the other, are understating temperature change. I would suspect that the Medieval Warm Period of 900 AD to 1200 AD was substantially warmer than the graphs show, and the Little Ice Age of 1600 AD to 1800 AD was substantially colder. If the graphs showed this, then the 'hockey stick' surge in the Mann graph of 1910 to 1998 would appear to be within normal variation.

 

Let me show you a further graph.

 

http://putfile.com/pic.php?pic=6/16504483939.jpg&s=f5

 

This shows the average results for 169 glaciers shrinking over the last several hundred years (since the Little Ice Age) thus representing global warming. You will note that there is little change in shrinkage rate from 1810 to 2000 AD. This implies that, on average, the trend in temperature rise over that time

was almost linear. This ties in with my interpretation of the Mann graph - indicating that the 'hockey stick' surge was an artifact of the change in measurement method.

[swansont]

Swansont.

Both graphs vary by no more than 0.2C from 1000 AD to 1890. The Mann graph then shows a surge of 0.7C.

 

Rubbish. Your second graph doesn't have a temperature scale. It is in units of mm. Without a calibration' date=' you can't say anything about the temperature range present in the lower graph.

 

Again, you are avoiding the main point. And that is that the Mann graph shows a major change in results 'coinciding' with a change in measurement technique, while the other graph, which keeps the same measurement method, shows no significant change at that point. I know you get the point, and are just raising quibbles to deny it to yourself.

 

Surely you know that correlation doesn't mean causality; it insufficient that the method changed at that point. Again, without knowing how tree-ring growth is related to temperature and other environmental parameters, none of the conclusions you have presented are valid. Is it linear? What if it's quadratic, or exponential? What if it also depended on, oh, rainfall, for example? Is it a local measurement, or a global one? (local areas can show different trends than the average). None of those factors are addressed, and they are all relevant.

 

Without that information, you can't draw any conclusion about what the graph means, other than what it says: tree growth (for whatever data sample was used) has shown variation over time.

 

 

Your concern about the Mann graph is valid. A change in method is something that should be investigated. But you haven't presented a scientifically valid contradiction to it. Did Mann address that concern in the original literature? I know of other experiments where the dynamic range of an apparatus is insufficient to collect all the data you want, so that multiple methods have to be used; it's no problem if you calibrate your detectors, so it's not something that automatically invalidates the measurement.

[bascule]

the Mann graph shows a major change in results 'coinciding' with a change in measurement technique

 

It isn't a "change" in measurement technique. It's the introduction of a new measurement technique, and the first one which has any degree (pun unintentional) of accuracy or precision in terms of assessing the temperature.

 

Are you trying to suggest that the GCM inputs are being skewed by bad data? If not, the GCMs agree with empirical measurements, and you have no argument whatsoever. If you are, you'll find information about the input data used by the 12 different GCMs in this graph in the papers for their corresponding studies:

 

 

If you're going to argue against the validity of the GCM input data, look up a paper, find what inputs they're using, and argue against that.

 

Otherwise, your arguments are completely unfounded.

[Dr. Dalek]

Wow Deja Vu,

 

I'd like to point out that none of us got any where arguing In "Accuracy of an Inconvinent Truth", do you think this thred will be any different?

[SkepticLance]

Dr. Dalek.

You are undoubtedly correct. However, it is so much fun winding up people who have a totally arrogant belief in their own rightness.

 

bascule.

Thank you for reintroducing that graph. It illustrates so well the inaccuracies of the various study methods that I described, and the errors that run from one study to the next.

 

Swansont said :

 

Rubbish. Your second graph doesn't have a temperature scale. It is in units of mm. Without a calibration, you can't say anything about the temperature range present in the lower graph.

 

The units had been calculated in degrees also. Sorry, I kept the graph but not the article, so you will have to take my word for it, or not as you please. However, the pre 1890 part of the graph fit smack in the middle of the range of results shown in bascules graph. It is only after that, where the measurement method changed, that a dramatic difference appeared.

 

Swansont, you failed to comment on the glacier shrinkage graph, which ties in perfectly with my argument. Do you not have an answer?

[bascule]

Thank you for reintroducing that graph. It illustrates so well the inaccuracies of the various study methods that I described, and the errors that run from one study to the next.

 

I guess it's cut-n-paste argument time again:

 

Are you trying to suggest that the GCM inputs are being skewed by bad data? If not' date=' the GCMs agree with empirical measurements, and you have no argument whatsoever. If you are, you'll find information about the input data used by the 12 different GCMs in this graph in the papers for their corresponding studies:

 

[ We've all seen this now ']

 

If you're going to argue against the validity of the GCM input data, look up a paper, find what inputs they're using, and argue against that.

 

Otherwise, your arguments are completely unfounded.

[timo]

You need to think more out of the box, Bascule.

[swansont]

Dr. Dalek.

You are undoubtedly correct. However' date=' it is so much fun winding up people who have a totally arrogant belief in their own rightness.

 

...

 

Swansont said :

 

[b']Rubbish. Your second graph doesn't have a temperature scale. It is in units of mm. Without a calibration, you can't say anything about the temperature range present in the lower graph.[/b]

 

The units had been calculated in degrees also. Sorry, I kept the graph but not the article, so you will have to take my word for it, or not as you please. However, the pre 1890 part of the graph fit smack in the middle of the range of results shown in bascules graph. It is only after that, where the measurement method changed, that a dramatic difference appeared.

 

Swansont, you failed to comment on the glacier shrinkage graph, which ties in perfectly with my argument. Do you not have an answer?

 

 

So where, exactly, am I wrong? You haven't provided a scientifically valid criticism. You have only provided a graph of tree growth measurements. And the temperature calibration wasn't the only limitation I mentioned. I can get (I provided a link before, in the other thread) the temperature data for regions of the US that show that parts of it have cooled over the last 100 years. Does that prove that world hasn't warmed up? So local vs global has to be addressed as well.

 

It's too bad you see this as arrogance. The objection you have presented, scientifically, is crap. That doesn't make the Mann curve right, but I've never argued that point, anyway.

 

Your glacier graph was way too small to read. Doesn't it exist online somewhere else?