Ice cores have proven to be an invaluable resource of information about past climate, showing cyclical ice ages, natural disasters and countless examples of abrupt climate change as far back as 800 thousand years ago. You can watch a short video about Ice core research on the US National Science Foundation Ice Core Facility website.
Source: Image by Michael Sigl, Frozen in time article, Royal Society of Chemistry.
Ice cores are frequently discussed in conversations and articles about climate change, but the data presented is always from the ancient past, thousands of years ago. Yet most of the Ice that has been drilled is from the recent past, covering the modern period of global warming. So where is the data and what does it say about global warming?
Ice core temperature records
Before we examine the data, lets see how ice cores work as a temperature record. It actually has nothing to do with the rings visible in the core, these are caused by particulates like pollen and ash present in the ice. It is the isotopic composition of the water which is used as a temperature proxy (a parameter which correlates temperature).
Source: Image by Heidi Roop, 'Core questions: An introduction to ice cores' article, NASA.
Water is a molecule made up of two elements Oxygen and Hydrogen, both of which come in a range of isotopes (heavier or lighter versions of themselves). It turns out that the ratio of the heavier isotope to the lighter isotope correlates well the temperature at the time the snow fell. The reason for this is pretty simple to understand.
Ice sheets are formed at high altitudes (mountains) and high latitudes (the poles). It requires energy to lift and transport water vapour from the ocean surface to these locations. When the climate system is cooler, the planets weather cells are weaker so the heavier isotopes tend to fall as rain before reaching the ice sheets, causing the percentage of heavy molecules in the snow falling on the ice to reduce. On the other hand when the climate is warmer and the weather cells are stronger, the percentage increases. Simple.
Delta 18-Oxygen ratio
Oxygen is the element most commonly used as a temperature proxy, by measuring the ratio of the heavier isotope, having 18 nucleons (neutrons and protons), to the lighter isotope, having 16 nucleons. This is compared to the same ratio for a precise and well known standard and calculated as a percentage difference; read more about this here.
The delta 18-Oxygen ratio, or δ18O for short, correlates well with temperature. In rainfall, a 60°C change in temperature corresponds to a 4% change in the Oxygen isotope ratio. In Ice records it is more common to see this ratio expressed in per-mille ‰ (per thousand), rather than per-cent % (per hundred), making the conversion factors 0.67‰ per 1°C, or 1.5°C per 1‰. Read more about this relationship in this article from NASA.
Source: Reibeek, H. NASA Earth Observatory.
Long term variations of δ18O in ice cores from Greenland show the emergence of the northern hemisphere from a cold and surprisingly chaotic period of glaciation, the last Ice Age, into the present warm and comparatively stable period called the Holocene. The change in the average level of δ18O from the ice age to the holocene was 5‰ (five per-mille), which corresponds to an average temperature increase of 7.5°C in Greenland.
Source: Schulz and Paul.
The signature of Global Warming
If the climate is warming, just as the ice cores show the warming at the end of the Ice Age, they will show the warming over the past century or so. The amount the climate has warmed is called the global Annual Temperature Anomaly (ATA), which is a fancy way of saying the change relative to the average temperature during a particular period.
The Annual Temperature Anomaly
There are actually three different centres calculating average global temperatures, which work independently and use different methods for collecting and processing data (Met Office). As such there are three slightly different data sets for the Annual Temperature Anomaly (ATA). Essentially they take temperature measurements from weather stations around the world and average them together in a way which they think produces a value representative of the entire planet.
The curve below is produced by the Goddard Institute for Space at NASA, which shows the annual temperature anomaly relative to the average temperature between 1951 and 1980. In 2020 the anomaly is 1°C, and the total amount of warming since 1900 is about 1.5°C.
Source: NASA's Goddard Institute for Space Studies (GISS).
Corresponding Oxygen Isotope levels
In the ice cores, a 1.5°C temperature increase since 1900 would appear as an increase in δ18O isotope levels of 1‰. Applying the conversion factor of 0.67‰ per 1°C, we can turn the temperature anomaly into a δ18O anomaly that we can look for in the ice records.
Since warming is a trend, we need to look at the rate of change of temperature, which is the gradient of a straight line fitted to the data. Rather than fit a line through the whole period from 1900 to 2019, we will break it into two 60 year periods either side of 1960.
Before 1960 the gradient is +0.00463‰ per year, or +0.463‰ per century.
After 1960 the gradient increases to +0.01136‰ per year, or +1.136‰ per century.
The change in gradient is +0.00673‰ per year, or +0.673‰ per century.
We can refer to all of these trends simply as +1 ATA, that is one times the Annual Temperature Anomaly (ATA). This is valid for all three values, the gradients for the periods before and after 1960 and the change in gradient between them. In the analysis that follows, this data would be presented as:
1900-1959 +1.0 ATA. 1960-2019 +1.0 ATA. Change +1.0 ATA. Increased warming.
Moderate warming prior to 1960, followed by "Increased warming" post 1960 is characteristic of the Annual Temperature Anomaly. This is the signature of Global Warming and what we will be looking for in the ice core records.
Arctic amplification
The rate of warming should be even higher for ice cores from the Arctic region due to an effect known as "Arctic amplification". The map below shows the temperature anomaly for the period from 2000 to 2009. Over this period the warming in the arctic was 3.3 times the global average. Read more about this on this webpage from NASA.
Source: Robert Simmon, NASA.
Accounting for this effect we should see trends in some of the ice core data, particularly from cores in the Arctic or West Antarctic Peninsula as high as 3.3 times the global trends in the Annual Temperature Anomaly, which works out at 1.53‰ per century before 1960 and 3.75‰ per century after 1960. We will refer to these amplified trends as +3.3 ATA. In the analysis that follows, this data would be presented as:
1900-1959 +3.3 ATA. 1960-2019 +3.3 ATA. Change +3.3 ATA. Increased warming.
Anticipated warming trends
According to the map by NASA, most of the world is warming, with very few areas cooling. Therefore we should expect to see warming trends in the vast majority of ice core data, being anywhere between +1 ATA and +3.3 ATA.
In the absence of long term heat transfer between regions around the globe, we should expect to see the same multiple of ATA for the gradient before 1960, the gradient after 1960 and also the change in gradient. For example a site somewhere in the Arctic might be warming at twice the ATA, in which case we would expect:
1900-1959 +2.0 ATA. 1960-2019 +2.0 ATA. Change +2.0 ATA. Increased warming.
Accounting for distortion by long term heat transfer
In the presence of long term heat transfer, that is a linear trend occurring over hundreds or even thousands of years as heat is moved to or from a given place on earth, the gradients before and after 1960 will be altered, but the change in gradient will not.
For example if a long term cooling trend due to heat transfer of -1‰ per century (-1‰pc), was added to the +2 ATA warming trends above, the result would be:
1900-1959 -1‰pc + 2*(+0.463‰pc) = -0.074‰pc = -0.2 ATA. Cooling.
1960-2019 -1‰pc + 2*(+1.136‰pc) = +1.272‰pc = +1.1 ATA. Warming.
Change +1.272‰pc - -0.074‰pc = +1.346‰pc = +2.0 ATA. Increase.
The gradient increased, transition from cooling to warming. Summary "Warming".
On the other hand if it were a long term warming trend due to heat transfer:
1900-1959 +1‰pc + 2*(+0.463‰pc) = +1.926‰pc = +4.2 ATA. Warming.
1960-2019 +1‰pc + 2*(+1.136‰pc) = +3.272‰pc = +2.9 ATA. Warming.
Change +3.272‰pc - +1.926‰pc = +1.346‰pc = +2.0 ATA. Increase.
The gradient increased, exaggerating the rate of warming. Summary "Increased warming".
So in interpreting results, we need to be aware that long term heat transfer between locations around the world will alter the trends, with the period before 1960 being the most sensitive to it, the period after 1960 being less sensitive, and the change in gradient between the two being unaffected.
Do Ice Cores show the signature of global warming?
Step 1 - Find and prepare the Ice Core data
There are many publications publishing and discussing the findings of Ice Core measurements. However for what we want to attempt, we need to source the original data used for these publications. Brilliantly the data for the majority, if not all, Ice Cores is available to download from the National Centres for Environmental Data at NOAA. Click the picture below to visit their website, where you can see a full list of Ice Core datasets by title and investigator.
All of the data used in this analysis was sourced from the above website. For each of the datasets considered below, the name contains a link to the relevant page on NOAA's website where you can download the original data for yourself.
To prepare the data we simply need to isolate the Oxygen isotope data, and divide it into two periods before and after 1960.
Step 2 - Categorise datasets by trends
A trend can be positive, negative or in rare cases exactly zero. Since none of the datasets examined had a trend which was exactly zero, this leaves us with the following categories:
Increased warming - warming before 1960, warming at faster rate after 1960.
Decreased warming - warming before 1960, warming at slower rate after 1960.
Warming - cooling before 1960, warming after 1960.
Cooling - warming before 1960, cooling after 1960.
Decreased cooling - cooling before 1960, cooling at slower rate after 1960.
Increased cooling - cooling before 1960, cooling at faster rate after 1960.
Step 3 - Eliminate datasets where the change in trend is negative
The Annual Temperature Anomaly has a step increase in the gradient of the trend after 1960. As we saw previously, whilst the gradient of the trends either side of 1960 are changed by the addition of long term trend, the change in gradient between them is not. Therefore we can eliminate any of the categories above which do not contain a similar change in gradient.
The only categories with a similar increase in gradient after 1960 are "increased warming", "warming" and "decreased cooling". A dataset could fall into one of these categories if the ATA was added to a positive trend, a small negative trend, or a large negative trend respectively.
Antarctica
Amundsenisen, Dronning Maud Land, Antarctica 75°S 6°E
DML01: 1900-1959 -3.3 ATA. 1960-1995 -6.4 ATA. Change -8.5 ATA. Increased cooling.
DML04: 1905-1959 -1.2 ATA. 1960-1995 +2.1 ATA. Change +4.3 ATA. Warming.
DML06: 1900-1959 -0.5 ATA. 1960-1996 -0.3 ATA. Change -0.2 ATA. Increased cooling.
DML08: 1919-1959 -2.5 ATA. 1960-1996 +7.6 ATA. Change +14.6 ATA. Warming.
DML09: 1900-1959 +17.1 ATA. 1960-1995 +6.4 ATA. Change -1.0 ATA. Decreased warming.
DML10: 1900-1959 +7.8 ATA. 1960-1996 +0.4 ATA. Change -4.8 ATA. Decreased warming.
Dronning, Dronning Maud Land, Antarctica 71°S 12°E
1902-1959 -5.1 ATA. 1960-2006 +2.8 ATA. Change +8.2 ATA. Warming.
Gomez, Antarctic Peninsula, Antarctica 74°S 70°W
1900-1959 +0.8 ATA. 1960-2005 +2.1 ATA. Change +3.1 ATA. Increased warming.
Canada
Agassiz, Ellesmere Island, Canada 81°N 73°W
A77: 1900-1959 +0.9 ATA. 1960-1977 +0.2 ATA. Change -0.2 ATA. Decreased warming.
A84: 1900-1959 +3.6 ATA. 1960-1973 -3.6 ATA. Change -8.5 ATA. Cooling.
Greenland
GISP2, Summit, Greenland 73°N 39°W
B: 1900-1959 +2.6 ATA. 1960-1986 -6.8 ATA. Change -13.3 ATA. Cooling.
GRIP, Summit, Greenland 73°N 39°W
891: 1900-1959 +0.8 ATA. 1960-1986 +1.3 ATA. Change +1.6 ATA. Increased warming.
892: 1900-1959 +1.6 ATA. 1960-1986 -2.8 ATA. Change -5.8 ATA. Cooling.
912: 1900-1959 +2.0 ATA. 1960-1986 -0.4 ATA. Change -2.1 ATA. Cooling.
South America
Quelccaya, Peru 14°S 71°W
1900-1959 +1.0 ATA. 1960-1986 +1.6 ATA. Change +1.9 ATA. Increased warming.
Tibet
Dasuopu, Tibet, China 28.38°N 85.72°E
Core 3: 1900-1959 +7.1 ATA. 1960-1986 +1.0 ATA. Change -3.2 ATA. Decreased warming.
Table of results
For convenience I've tabulated the results for all 16 of the datasets analysed below, with the columns being the trend before 1960, after 1960 and the change, all relative to ATA values.
Trends: before 1960 | after 1960 | change
Amundsenisen DML01: -3.3 ATA | -6.4 ATA | -8.5 ATA. Increased cooling.
Amundsenisen DML04: -1.2 ATA | +2.1 ATA | +4.3 ATA. Warming.
Amundsenisen DML06: -0.5 ATA | -0.3 ATA | -0.2 ATA. Increased cooling.
Amundsenisen DML08: -2.5 ATA | +7.6 ATA | +14.6 ATA. Warming.
Amundsenisen DML09: +17.1 ATA | +6.4 ATA | -1.0 ATA. Decreased warming.
Amundsenisen DML10: +7.8 ATA | +0.4 ATA | -4.8 ATA. Decreased warming.
Dronning: -5.1 ATA | +2.8 ATA | +8.2 ATA. Warming.
Gomez: +0.8 ATA | +2.1 ATA | +3.1 ATA. Increased warming.
Agassiz A77: +0.9 ATA | +0.2 ATA | -0.2 ATA. Decreased warming.
Agassiz A84: +3.6 ATA | -3.6 ATA | -8.5 ATA. Cooling.
GISP2 B: +2.6 ATA | -6.8 ATA | -13.3 ATA. Cooling.
GRIP 891: +0.8 ATA | +1.3 ATA | +1.6 ATA. Increased warming.
GRIP 892: +1.6 ATA | -2.8 ATA | -5.8 ATA. Cooling.
GRIP 912: +2.0 ATA | -0.4 ATA | -2.1 ATA. Cooling.
Quelccaya: +1.0 ATA | +1.6 ATA | +1.9 ATA. Increased warming.
Dasuopu Core 3: +7.1 ATA | +1.0 ATA |-3.2 ATA. Decreased warming.
Increased warming: 3.
Decreased warming: 4.
Warming: 3.
Cooling: 4.
Decreased cooling: 0.
Increased cooling: 2.
Discussion
Six of the datasets displayed a positive change in the trend, and could therefore be considered as qualitatively consistent with the Annual Temperature Anomaly. By "qualitatively", I mean the data has the right characteristic shape, without consideration of the numerical values. These results are shown below:
Trends: before 1960 | after 1960 | change
Amundsenisen DML04: -1.2 ATA | +2.1 ATA | +4.3 ATA. Warming.
Amundsenisen DML08: -2.5 ATA | +7.6 ATA | +14.6 ATA. Warming.
Dronning: -5.1 ATA | +2.8 ATA | +8.2 ATA. Warming.
Gomez: +0.8 ATA | +2.1 ATA | +3.1 ATA. Increased warming.
GRIP 891: +0.8 ATA | +1.3 ATA | +1.6 ATA. Increased warming.
Quelccaya: +1.0 ATA | +1.6 ATA | +1.9 ATA. Increased warming.
None of them are a straightforward amplification of the ATA, in which case all three of the numbers would be similar.
Superposition of amplified ATA with long term regional cooling
Previously we considered a model based on a long term linear trend added to a multiple of the Annual Temperature Anomaly. In doing so we found that adding a negative trend caused a significant reduction of the trend before 1960, a slight reduction of the trend after 1960, and had no effect on the change. All of these datasets show that general behaviour.
If our model is correct, then the amount of warming at these locations would vary between between 1.6 and 14.6 times ATA.
Cherry picking
If these were the only datasets I presented, I might be able to convince you that this is a pretty decent model to account for regional temperatures at the locations of each of these Ice Sheets.
Leaving out Amundsenisen DML08 and Dronning would reduce the range of warming to between 1.6 and 4.3 ATA, which would be even more compelling as these sort of numbers sit better with those typically discussed in relation to Arctic Amplification.
But this is all blatant cherry picking. In reality there are 10 more datasets which don't fit that model at all, and these are just amongst the datasets that I've examined.
ATA is absent from most of the Ice Cores studied
In truth I only found traces of the ATA in 6/16 models, which is only 38%, and even these were only a "qualitative" match. Just because those datasets look more like the ATA, doesn't mean they are the ATA. The other 62% of models didn't match the characteristic stepped increase in gradient of the ATA, and some even went the other way entirely.
Ignoring ATA, most Ice Cores show warming
Ignoring the ATA and simply grouping all patterns of warming, or cooling together, we find that 10 datasets showed warming, compared to only 6 which showed cooling. So 63% of the Ice Cores examined show some level of warming after 1960.
However, it should be acknowledged that 4 of these datasets showed a greater rate of warming prior to 1960 than after 1960, which isn't compatible with global warming theory.
Additional dynamics and accuracy of the data
The Oxygen isotope data in each of the datasets is full of additional dynamics, many of which are greater in amplitude than the signal we are looking for. The peak-to-peak variation in the data is typically around 5‰, whereas the signal due to ATA is only a 1‰ rise over the whole 120 years combining both periods either side of 1960.
It is not possible simply from looking at the data to tell whether these variations are real or are simply measurement error. Whilst real Oxygen isotope levels do correlated with temperature, they may also be influenced by other factors. Additionally there may be errors in the accuracy of measuring isotope levels, or in how the core has been divided up into annual sections for measurement.
Conclusion
In this article I have examined the Oxygen Isotope data for quite a few of the Ice Cores spanning the 20th Century. I wanted to see how obvious global warming is in the measurements, and the answer is its not. Whilst a little over a third of the datasets examined show a stepped increase in gradient around 1960, characteristic of the Annual Temperature Anomaly, the majority of the results do not show this, and some even show a stepped decrease.
Most of the Oxygen Isotope datasets are full of dynamics greater in amplitude than the signal we are looking for. If these dynamics represent real variations in temperature (they may not), then we need to understand their cause and how frequently they occur, in order to distinguish more confidently what we believe to be global warming from natural variations.
Comments