Global Warming - Icecore Data
Past Temperatures

Ice cores have been collected from around the world. Stored in these cores are various physical properties which are interpreted to reveal information about the climate at the time the snow fell. This page presents these methods used to determine past temperatures and shows that they do not agree.

These plots show, without a doubt, that none of the "temperature proxies" agree with each other. (These are from my Vostok and Dye-3 pages.)

Vostok Deuterium and Temperature Vostok Deuterium and Temperature Dye-3 δ18O and Borehole Temperature
Deuterium (blue), delta T (red) Deuterium (blue), delta T (red) - All δ18O (red), delta T (black)

Deuterium and δ18O - A Deuterium and δ18O - B Deuterium and δ18O - C
Full Core Magnified - 0 to 500 meters Magnified - 1,500 to 2,000 meters
Blue: Deuterium, Orange: δ18O

In the first plot, the change in the deuterium concentration was "adjusted" to account for changes in the amount of deuterium in the oceans. This actually makes sense. However, a close inspection of the plots indicates that the adjustments are made in distinct steps. The only paper I have found that explains exactly what was done is behind a pay wall. The first plot shows the last 11,000 years with the curves adjusted to show what was changed. The second graph shows all the data without any adjustments (except for automatic scaling for two y-axes). I strongly suggest that you plot the data yourself so that you can zoom in to see the problems.

The actual changes are not very significant .. unless you are trying to use this data to prove anthropogenic global warming (AGW). In that case, converting a 11,000 year 1.5K temperature decrease into no change here which is, of course, used to prove that the current 0.5K increase must be due to people.

The Dye-3 plot shows the complete lack of agreement between δ18O and reconstructed temperatures based on actual borehole temperature measurements. Of course, the scales can be changed so that the black line is flat, but that is not supported by other data.

One of the things that bothers me is that deuterium is reported for the Antarctica cores and δ18O is reported for the Greenland cores. I was fortunate to find an alternate source for the Vostok δ18O data .. which is what I used to create the bottom 3 graphs. In my opinion, these graphs prove, without a doubt, that there is no correlation between temperature and δ18O.


Deuterium is a naturally occurring isotope of hydrogen. As a result, it is found in water (ice). Because of the small difference in mass, the ratio of deuterium to hydrogen in sea water is slightly different than the ratio in rain. There is significant evidence that this ratio is related to temperature.

However, there are several obvious problems with this

And, there is the other problem - once snow falls, some of it sublimes (evaporates without first melting). Of course, the lighter molecules are more likely to sublime. This is important - because of the way snow is changed to ice, it may be exposed to over 1,000 years of sublimation. Basically, the metamorphosis (conversion of snow to ice) is not complete until there are 70 to 100 meters of accumulation. Thus, the deuterium concentration depends on how much snow falls each year.

There is, of course, a 100% correlation - when it is warmer

Since more sublimation should cause the deuterium concentration to increase, the fact that the deuterium peaks occur when there are the minimum number of years to convert snow to ice indicates that sublimation causes the peaks to be lower and the troughs to be higher.

To be clear, at Vostok, it takes about 2,000 years for snow to become ice during the warm interglacials and over 6,500 years during the cold glacial maxima.


In Antarctica, deuterium is used as the primary temperature proxy. In Greenland, δ18O is used. In theory, δ18O is a better choice because it is heavier. However, there appears to be a lot less 18O than deuterium. In addition, oxygen is trapped both as water (ice) and as a gas (O2). Most of the reports I've seen try to determine δ18O in the gas. However, there does not appear to be a correction for the fact that, apparently, the O2/N2 ratio in ice cores is different than what is found in the atmosphere. (This is a serious problem that needs resolution - whatever is causing the O2/N2 ratio change is probably also affecting all the other trapped gases .. including CO2.)

I only have a few datasets that permit the plotting of δ18O with other temperature proxies - they don't agree. Specifically, deuterium indicates that there are 4 interglacials recorded in the Vostok ice core - δ18O indicates 8 (eight).

In coral, lake bottom, and sea bottom cores, δ18O is the primary temperature indicator via CaCO3. However, this reference clearly states that δ18O is a salinity indicator.

Borehole measurements

Get this - you can simply drop a thermometer down an empty borehole and measure the temperature of the surrounding ice. (Wow, that sounds simple!)

There are, of course, a couple of "complications". The first is that the hole is filled with a working fluid (typically, an oil) when the hole is drilled. Also, the act of drilling a hole produces heat. In addition, heat flows in the ice and causes the highs and lows to average out.

Several researchers have tried to develop algorithms to determine a surface temperature profile which could have produced the current readings. The main problem with this is that the solution is not unique - there are many profiles that could have produced the current readings.

As a result, these reconstructions should not be used beyond their limitations. However, they give good information when the temperature is steady for ten's of thousands of years.

The Dye-3 plot above shows about 4,000 years of data. It is pretty obvious that there is no correlation between the reconstructed borehole temperatures and the δ18O proxy. However, there is a very close correlation between the plot above and non-Ice_Boreholes.

Of course, none of these proxies can be used without calibration and adjustments. When skeptics request data, full explanations of these adjustments are also necessary.

Diffusion: The limit to δ18O-based ice core dating describes using the annual δ18O variations to date ice cores. Other pages discuss various properties of the ice cores. Most pages at this site contain an excellent flash tool that shows the locations of the various Greenland boreholes - be sure to click on each site for more information.

Author: Robert Clemenzi
URL: http:// / Global_Warming / Icecore_Data / Temperatures.html