The Vostok ice core (located at 7827'51.92"S 10650'14.38"E in Antarctica) reached a depth of 3,623 meters. Below 3,311 meters there are significant discontinuities in the data. As a result, it is assumed that the ice below that point has been disturbed. Most likely, this means that the interglacial temperature was warm enough that surface melting occurred. This would be possible even if the planet was no warmer than today because the snow was closer to sea level. With an environmental lapse rate of 6.5K per kilometer, that means that the surface temperature could have been about 21.5K warmer than the current high .. less if you account for the fact that Antarctica is sinking under the weight of the ice. However, there are other possibilities - for instance, that point may have been part of a moving glacier. And, of course, there is always the possibility of meteor impact, volcanoes and other factors.

Since the South Pole currently "warms" to temperatures around -17C, it is very likely that thinner snow cover could have allowed temperatures above freezing.

δ Deuterium and δ Temperature

It is normally assumed that, like d18O, the change in deuterium with respect to the ocean value can be used as a proxy for temperature. However, there is an anomaly in the Vostok data - as shown below.

This is a series of graphs showing Deuterium (blue) and delta T (red) from deutnat.txt. According to the readme text, there is a linear relationship between the deuterium and temperature values .. adjusted for the change in deuterium concentration in the ocean. Unfortunately, I have not been able to read the original references (they cost $40 or more each just to read them), so I have assumed a constant oceanic deuterium concentration. In these graphs, I have shifted the deuterium curve (+10, +3, +1) and left the temperature curve (red) alone.

Deuterium + 10 Deuterium + 3 Deuterium + 1
Zero to 7,000 BP overlaps 25,000 to 30,000 BP overlaps 11,000 to 20,000 BP overlaps
The relative gain of the two curves is fairly arbitrary just to make them both fit on the same graph and to make the minimum and maximum values on the two vertical axes integers. I did play with the gains to see if the lack of fit was simply due to the difference in scaling ... it is not. In fact, the third plot (Deu+1) proves it because there is an extremely good overlap over the complete range from 11,000 to 20,000 BP.

Below is the same data, but from today back to 11,000 BP - notice that the shape of the two lines is quite different. Simple scaling can not account for this. This plot clearly shows that there is almost a 1K difference between the "official" temperature and what I would have guessed at about 11,000 years ago. However, and most important, the temperature plot shows no change over 11,000 years and the deuterium plot suggested that there was actually a 1.5K decrease over the same time period. (In my opinion, this is a big deal. Before accepting any "Global Warming" theory, I would want this difference explained .. in detail .. and not at $40 per paper.)

Deuterium + 10 Deuterium + 11
Zero to 7,000 BP overlaps Magnified

(The reason one plot is D+10 and the other is D+11 is related to the change in vertical gain. Remember, I selected fairly arbitrary gains just to make the data fit. The point is that the data fits almost exactly for over 50% of the data points and then diverges significantly.)

According to the references, the relative deuterium values have a measurement accuracy of 0.5/ Surface Mean Ocean Water (SMOW).

Remember, as with all my plots, you can download the data and plot these for yourself. If you like, you can download and use my plotting program, or use another program - MatLab, Excel, whatever you like. The data is a simple tab delimited text file. If you do, you will see that this convergence and divergence of the two plots goes on throughout the 420,000 years covered. I have not shown the details here because there would have to be too many plots to see the details.

Note: Apparently, the important deuterium data between 312 and 320 meters below surface (mbs) (the warming just before the Younger Dryas) is invented from someone's imagination .. according to the notes in deutnat.txt. This is not obvious because of the "fabricated ?" noise in the data.

Deuterium and δ18O

Both Deuterium and δ18O are supposed to be proxies for temperature. However, for the same core, the data does not agree. Using nsidc data the following graphs show the differences. Unfortunately, the nsidc data only has depth associated with it. Oh, some age data is available, but I will have to write yet another program to use it because there are several different models used to get age (which is actually a good thing) and because the age mappings are in separate files from the δ18O data. (Blue: Deuterium, Orange: δ18O)

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

While there is good general agreement, there are several very significant differences. To begin with, the δ18O peaks show many shorter ice ages instead of the 4 typically discussed in the literature. This is a major problem because the 4 typically acknowledged cycles associated with changes in deuterium almost correlate with the Milankovitch cycles while the 11 or so δ18O peaks (which are clearly visible) disprove the standard theory (or prove that δ18O is not a temperature proxy).

I have magnified two sections to illustrate other differences. Image B (above) shows the most recent period, the top 500 meters or about 24,000 years. The timing of the Younger Dryas (at 300 meters) is very troublesome because δ18O clearly shows the start about 1,000 years before the deuterium signature. The logical interpretation is that this is simply an "age of ice" verses "age of air" issue .. except that adjusting for that after the Younger Dryas means that the curves do not align before the cooling (340 to 350 meters). Also, the significantly higher temperature suggestion at about 10,000 BP, when combined with the analysis above, indicates that temperatures may have been very much higher than any paper has suggested.

Image C shows a more significant 50 meter timing difference at around 1,900 meters (which corresponds to the termination of the previous ice age about 133,000 BP). In this case, adjusting for the ice/air difference makes the gap wider (because the air is always newer than the ice it is trapped in).

Unfortunately, this data is for the δ18O of the trapped air, and not for the δ18O of the ice itself. As a result, there will be a significant difference between the deuterium and δ18O changes. However, the fact that δ18O changes after the deuterium signature (at 1,900 meters) suggests that there is a major problem using these isotopes as temperature proxies.

Actually, the δ18O data is obtained from several nearby (same site) cores and then combined into a single file (by them, not me). Also, there are actually two separate columns of data - δ18O and δ18O Atm. The associated papers explain that δ18O Atm is the better indicator of temperature because it has been adjusted for the rate of ice accumulation. However, for the discussion above, the differences are not significant.

As always, I strongly suggest that you plot this data yourself. The number of obvious, and extremely large, differences should easily convince you that more research is needed and/or that all paleoclimate temperature reconstructions are basically worthless.

Author: Robert Clemenzi
URL: http:// questionable-science.com / Global_Warming / Icecore_Data / Vostok.html