Ice core basics

Why use ice cores? | How do ice cores work? | Layers in the ice | Information from ice cores | Further reading | References | Comments |

Why use ice cores?

420,000 years of ice core data from Vostok, Antarctica research station. Current period is at left. From bottom to top: * Solar variation at 65°N due to en:Milankovitch cycles (connected to 18O). * 18O isotope of oxygen. * Levels of methane (CH4). * Relative temperature. * Levels of carbon dioxide (CO2). From top to bottom: * Levels of carbon dioxide (CO2). * Relative temperature. * Levels of methane (CH4). * 18O isotope of oxygen. * Solar variation at 65°N due to en:Milankovitch cycles (connected to 18O). Wikimedia Commons.
420,000 years of ice core data from Vostok, Antarctica research station. Current period is at right. From bottom to top: * Solar variation at 65°N due to en:Milankovitch cycles (connected to 18O). * 18O isotope of oxygen. * Levels of methane (CH4). * Relative temperature. * Levels of carbon dioxide (CO2). From top to bottom: * Levels of carbon dioxide (CO2). * Relative temperature. * Levels of methane (CH4). * 18O isotope of oxygen. * Solar variation at 65°N due to en:Milankovitch cycles (connected to 18O). Wikimedia Commons.

Ice sheets have one particularly special property. They allow us to go back in time and to sample accumulation, air temperature and air chemistry from another time[1]. Ice core records allow us to generate continuous reconstructions of past climate, going back at least 800,000 years[2]. By looking at past concentrations of greenhouse gasses in layers in ice cores, scientists can calculate how modern amounts of carbon dioxide and methane compare to those of the past, and, essentially, compare past concentrations of greenhouse gasses to temperature.

Ice coring has been around since the 1950s. Ice cores have been drilled in ice sheets worldwide, but notably in Greenland[3] and Antarctica[4, 5]. High rates of snow accumulation provide excellent time resolution, and bubbles in the ice core preserve actual samples of the world’s ancient atmosphere[6]. Through analysis of ice cores, scientists learn about glacial-interglacial cycles, changing atmospheric carbon dioxide levels, and climate stability over the last 10,000 years. Many ice cores have been drilled in Antarctica.

How do ice cores work?

This schematic cross section of an ice sheet shows an ideal drilling site at the centre of the polar plateau near the ice divide, with ice flowing away from the ice divide in all direction. From: Snowball Earth.
This schematic cross section of an ice sheet shows an ideal drilling site at the centre of the polar plateau near the ice divide, with ice flowing away from the ice divide in all direction. From: Snowball Earth.

The large Greenland and Antarctic ice sheets have huge, high plateaux where snow accumulates in an ordered fashion. Slow ice flow at the centre of these ice sheets (near the ice divide) means that the stratigraphy of the snow and ice is preserved. Drilling a vertical hole through this ice involves a serious effort involving many scientists and technicians, and usually involves a static field camp for a prolonged period of time.

Shallow ice cores (100-200 m long) are easier to collect and can cover up to a few hundred years of accumulation, depending on accumulation rates. Deeper cores require more equipment, and the borehole must be filled with drill fluid to keep it open. The drill fluid used is normally a petroleum-derived liquid like kerosene. It must have a suitable freezing point and viscosity. Collecting the deepest ice cores (up to 3000 m) requires a (semi)permanent scientific camp and a long, multi-year campaign[6].

Layers in the ice

If we want to reconstruct past air temperatures, one of the most critical parameters is the age of the ice being analysed. Fortunately, ice cores preserve annual layers, making it simple to date the ice. Seasonal differences in the snow properties create layers – just like rings in trees. Unfortunately, annual layers become harder to see deeper in the ice core. Other ways of dating ice cores include geochemisty, layers of ash (tephra), electrical conductivity, and using numerical flow models to understand age-depth relationships.

This 19 cm long of GISP2 ice core from 1855 m depth shows annual layers in the ice. This section contains 11 annual layers with summer layers (arrowed) sandwiched between darker winter layers. From the US National Oceanic and Atmospheric Administration, Wikimedia Commons.
This 19 cm long of GISP2 ice core from 1855 m depth shows annual layers in the ice. This section contains 11 annual layers with summer layers (arrowed) sandwiched between darker winter layers. From the US National Oceanic and Atmospheric Administration, Wikimedia Commons.

Although radiometric dating of ice cores has been difficult, Uranium has been used to date the Dome C ice core from Antarctica. Dust is present in ice cores, and it contains Uranium. The decay of 238U to 234U from dust in the ice matrix can be used to provide an additional core chronology[7].

Information from ice cores

Accumulation rate

The thickness of the annual layers in ice cores can be used to derive a precipitation rate (after correcting for thinning by glacier flow). Past precipitation rates are an important palaeoenvironmental indicator, often correlated to climate change, and it’s an essential parameter for many past climate studies or numerical glacier simulations.

Melt layers

Ice cores provide us with lots of information beyond bubbles of gas in the ice. For example, melt layers are related to summer temperatures. More melt layers indicate warmer summer air temperatures. Melt layers are formed when the surface snow melts, releasing water to percolate down through the snow pack. They form bubble-free ice layers, visible in the ice core. The distribution of melt layers through time is a function of the past climate, and has been used, for example, to show increased melting in the Twentieth Century around the NE Antarctic Peninsula[8].

Past air temperatures

It is possible to discern past air temperatures from ice cores. This can be related directly to concentrations of carbon dioxide, methane and other greenhouse gasses preserved in the ice. Snow precipitation over Antarctica is made mostly of H216O molecules (99.7%). There are also rarer stable isotopes: H218O (0.2%) and HD16O (0.03%) (D is Deuterium, or 2H)[9]. Isotopic concentrations are expressed in per mil δ units (δD and δ18O) with respect to Vienna Standard Mean Ocean Water (V-SMOW). Past precipitation can be used to reconstruct past palaeoclimatic temperatures. δD and δ18O is related to surface temperature at middle and high latitudes. The relationship is consistent and linear over Antarctica[9].

Snow falls over Antarctica and is slowly converted to ice. Stable isotopes of oxygen (Oxygen [16O, 18O] and hydrogen [D/H]) are trapped in the ice in ice cores. The stable isotopes are measured in ice through a mass spectrometer. Measuring changing concentrations of δD and δ18O through time in layers through an ice core provides a detailed record of temperature change, going back hundreds of thousands of years.

The figure above shows changes in ice temperature during the last several glacial-interglacial cycles and comparison to changes in global ice volume. The local temperature changes are from two sites in Antarctica and are derived from deuterium isotopic measurements. The bottom plot shows global ice volume derived from δ18O measurements on marine microfossils (benthic foraminifera) from a composite of globally distributed marine sediment cores. From Wikimedia Commons.
The figure above shows changes in ice temperature during the last several glacial-interglacial cycles and comparison to changes in global ice volume. The local temperature changes are from two sites in Antarctica and are derived from deuterium isotopic measurements. The bottom plot shows global ice volume derived from δ18O measurements on marine microfossils (benthic foraminifera) from a composite of globally distributed marine sediment cores. From Wikimedia Commons.

An example of using stable isotopes to reconstruct past air temperatures is a shallow ice core drilled in East Antarctica[10]. The presence of a “Little Ice Age”, a cooler period ending ~100 to 150 years ago, is contested in Antarctica. Disparate records often provide conflicting evidence. This ice core attempted to investigate the evidence for cooler temperatures during this period.

A 180 m deep ice core from the Ross Sea, Antarctica, was drilled by a team led by Nancy Bertler in 2001/2002[10]. The top 50 m of the ice core was analysed at 2.5 cm resolution using a continuous melting system. Ice core samples were analysed for stable isotope ratios, major ions and trace elements. An age model was extrapolated to the ice core using a firm decompaction model[10].  Deuterium data (δD) were used to reconstruct changes in summer temperature in the McMurdo Dry Valleys over the last 900 years. The study showed that there were three distinct periods: the Medieval Warm Period (1140 to 1287 AD), the Little Ice Age (1288 to 1807 AD) and the Modern Era (1808 to 2000 AD).

These data indicate that surface temperatures were around 2°C cooler during the Little Ice Age[10], with colder sea surface temperatures and possibly increased sea ice extent, stronger katabatic winds and decreased snow accumulation. The area was cooler and stormier.

Past greenhouse gasses

This photograph shows me (Bethan Davies) visiting Nancy Bertler and others in her ice core laboratory at GNS, New Zealand. The ice core is continuously melted and analysed by numerous automatic machines.
This photograph shows me (Bethan Davies) visiting Nancy Bertler and others in her ice core laboratory at GNS, New Zealand. The ice core is continuously melted and analysed by numerous automatic machines.

The most important property of ice cores is that they are a direct archive of past atmospheric gasses.  Air is trapped at the base of the firn layer, and when the compacted snow turns to ice, the air is trapped in bubbles. This transition normally occurs 50-100 m below the surface[6]. The offset between the age of the air and the age of the ice is accounted for with well-understood models of firn densification and gas trapping. The air bubbles are extracted by melting, crushing or grating the ice in a vacuum.

This method provides detailed records of carbon dioxide, methane and nitrous oxide going back over 650,000 years[6]. Ice core records globally agree on these levels, and they match instrumented measurements from the 1950s onwards, confirming their reliability. Carbon dioxide measurements from older ice in Greenland is less reliable, as meltwater layers have elevated carbon dioxide (CO2 is highly soluble in water). Older records of carbon dioxide are therefore best taken from Antarctic ice cores.

Other complexities in ice core science include thermal diffusion. Prior to becoming trapped in ice, air diffuses to the surface and back. There are two important fractionation processes: thermal diffusion and gravitational settling[11]. Thermal diffusion occurs if the surface is warmer or colder than the bottom boundary (the close-off depth). This temperature gradient occurs from climate change, which affects the surface first. The heavier components of the air (like stable isotopes) also tend to settle down (gravitational settling).

Thermal diffusion and gravitational settling can be measured and analysed because the fractionation of air follows well understood principles and relationships between different stable isotopes (namely, nitrogen and argon).

Other gasses

Other major gases trapped in ice cores (O2, N2 and Ar) are also interesting. The stable isotope concentration (δ18O) in ice core records mirrors that of the ocean. Oceanic δ18O is related to global ice volume. Variations of δ18O in O2 in ice core gasses are constant globally, making it a useful chronostratigraphic marker. It’s another way to relate ice-core chronologies.

Other ice-core uses

The vertical profile of an ice core gives information on the past surface temperature at that location[6].   In Greenland, glass shard layers from volcanic eruptions (tephra) are preserved in ice cores. The tephra ejected in each volcanic eruption has a unique geochemical signature, and large eruptions projecting tephra high into the atmosphere results in a very wide distribution of ash. These tephra layers are therefore independent maker horizons; geochemically identical tephra in two different ice cores indicate a time-synchronous event. They both relate to a single volcanic eruption. Tephra is therefore essential for correlating between ice cores, peat bogs, marine sediment cores, and anywhere else where tephra is preserved[12, 13].

Changes in sea ice concentrations can also be reconstructed from polar ice cores[14]. Ice core records of sea salt concentration reveal patterns of sea ice extent over longer (glacial-interglacial) timescales. Methane sulphonic acid in near-coastal ice cores can be used to reconstruct changes and interannual variability in ice cores.

Mineral dust accumulates in ice cores, and changing concentrations of dust and the source (provenance) of the dust can be used to estimate changes in atmospheric circulation[15]. The two EPICA ice cores (European Project for Ice Coring in Antarctica) contain a mineral dust flux record, showing dust emission changes from the dust source (glacial Patagonia). Changes in the dust emission is related to environmental changes in Patagonia.

Further reading


1.            Jouzel, J. and V. Masson-Delmotte, 2010. Deep ice cores: the need for going back in time. Quaternary Science Reviews, 29(27): 3683-3689.

2.            Augustin, L., C. Barbante, P.R.F. Barnes, J.M. Barnola, M. Bigler, E. Castellano, O. Cattani, J. Chappellaz, D. DahlJensen, B. Delmonte, G. Dreyfus, G. Durand, S. Falourd, H. Fischer, J. Fluckiger, M.E. Hansson, P. Huybrechts, R. Jugie, S.J. Johnsen, J. Jouzel, P. Kaufmann, J. Kipfstuhl, F. Lambert, V.Y. Lipenkov, G.V.C. Littot, A. Longinelli, R. Lorrain, V. Maggi, V. Masson-Delmotte, H. Miller, R. Mulvaney, J. Oerlemans, H. Oerter, G. Orombelli, F. Parrenin, D.A. Peel, J.R. Petit, D. Raynaud, C. Ritz, U. Ruth, J. Schwander, U. Siegenthaler, R. Souchez, B. Stauffer, J.P. Steffensen, B. Stenni, T.F. Stocker, I.E. Tabacco, R. Udisti, R.S.W. van de Wal, M. van den Broeke, J. Weiss, F. Wilhelms, J.G. Winther, E.W. Wolff, M. Zucchelli, and E.C. Members, 2004. Eight glacial cycles from an Antarctic ice core. Nature, 429(6992): 623-628.

3.            Johnsen, S.J., D. Dahl-Jensen, N. Gundestrup, J.P. Steffensen, H.B. Clausen, H. Miller, V. Masson-Delmotte, A.E. Sveinbjornsdottir, and J. White, 2001. Oxygen isotope and palaeotemperature records from six Greenland ice-core stations: Camp Century, Dye-3, GRIP, GISP2, Renland, and NorthGRIP. Journal of Quaternary Science, 16: 299-307.

4.            Mulvaney, R., N.J. Abram, R.C.A. Hindmarsh, C. Arrowsmith, L. Fleet, J. Triest, L.C. Sime, O. Alemany, and S. Foord, 2012. Recent Antarctic Peninsula warming relative to Holocene climate and ice-shelf history. Nature, 489: 141-144.

5.            Lambert, F., B. Delmonte, J.-R. Petit, M. Bigler, P.R. Kaufmann, M.A. Hutterli, T.F. Stocker, U. Ruth, J.r.P. Steffensen, and V. Maggi, 2008. Dust-climate couplings over the past 800,000 years from the EPICA Dome C ice core. Nature, 452(7187): 616-619.

6.            Brook, E.J., ICE CORE METHODS | Overview, in Encyclopedia of Quaternary Science, A.E. Scott, Editor. 2007, Elsevier: Oxford. 1145-1156.

7.            Aciego, S., B. Bourdon, J. Schwander, H. Baur, and A. Forieri, 2011. Toward a radiometric ice clock: uranium ages of the Dome C ice core. Quaternary Science Reviews, 30(19): 2389-2397.

8.            Abram, N.J., R. Mulvaney, E.W. Wolff, J. Triest, S. Kipfstuhl, L.D. Trusel, F. Vimeux, L. Fleet, and C. Arrowsmith, 2013. Acceleration of snow melt in an Antarctic Peninsula ice core during the twentieth century. Nature Geosci, advance online publication.

9.            Jouzel, J. and V. Masson-Delmotte, ICE CORE RECORDS | Antarctic Stable Isotopes, in Encyclopedia of Quaternary Science, A.E. Scott, Editor. 2007, Elsevier: Oxford. 1242-1250.

10.          Bertler, N.A.N., P.A. Mayewski, and L. Carter, 2011. Cold conditions in Antarctica during the Little Ice Age – implications for abrupt climate change mechanisms. Earth and Planetary Science Letters, 308: 41-51.

11.          Grachev, A.M., ICE CORE RECORDS | Thermal Diffusion Paleotemperature Records, in Encyclopedia of Quaternary Science, A.E. Scott, Editor. 2007, Elsevier: Oxford. 1280-1284.

12.          Abbott, P.M. and S.M. Davies, 2012. Volcanism and the Greenland ice-cores: the tephra record. Earth-Science Reviews, 115(3): 173-191.

13.          Hoek, W., Z. Yu, and J.J. Lowe, 2008. INTegration of Ice-core, MArine, and TErrestrial records (INTIMATE): refining the record of the Last Glacial – Interglacial Transition. Quaternary Science Reviews, 27(1): 1-5.

14.          Abram, N.J., E.W. Wolff, and M.A.J. Curran, 2013. A review of sea ice proxy information from polar ice cores. Quaternary Science Reviews, 79(0): 168-183.

15.          Fischer, H., F. Fundel, U. Ruth, B. Twarloh, A. Wegner, R. Udisti, S. Becagli, E. Castellano, A. Morganti, and M. Severi, 2007. Reconstruction of millennial changes in dust emission, transport and regional sea ice coverage using the deep EPICA ice cores from the Atlantic and Indian Ocean sector of Antarctica. Earth and Planetary Science Letters, 260(1): 340-354.

134 thoughts on “Ice core basics”

  1. “This method provides detailed records of carbon dioxide, methane and nitrous oxide going back over 650,000 years[6]. Ice core records globally agree on these levels, and they match instrumented measurements from the 1950s onwards, confirming their reliability.”

    I’m curious about this conclusion. How can you say that measurements correlating with instrumental data from 1950s onwards confirms reliability when you’re talking about samples that are thousands of years old? I know that you’re also saying that other ice core records also agree, but that just shows that ice core data gathered are consistent, but not necessarily accurate. We don’t really have a true control group from 400,000 years ago, do we? What kind of crushing and grinding does ice undergo over thousands of years?

    When bubbles are analyzed and you get your values, what exactly do those values represent? What about temperature, for instance? Does the data gathered represent the average for the entire year? What is the margin for error?

    What is the sampling rate for the graphs shown above? How many years of separation between the samples? There obviously aren’t 450,000 samples being used to create that graph, so I’m wondering how many samples there actually are. Could years of data have been lost to melting or sublimation?

    Thanks in advance for answering these questions.

    1. lynne morgan

      To match records from the 50’s you would use cores taken from places where snow accumulates quickly (in these places, the process of snow being compressed into more dense “firn” and eventually into ice takes a few decades) and compare this data to see if your interpretation of the weather, ascertained from that ice core, matches metreological records. And it does.

    2. I saw a report on 400,000 years of Vostok ice cores. They had a graph that also showed CO2. It amazed me that despite what their own graph showed (that higher CO2 FOLLOWED warming) they keep saying that warmer follows CO2.

      1. This is a common story spread by climate change deniers and their paid staff.

        What you need to do is stop and read some of the science.

        Of course the deniers will then tell you that every one of the hundreds of science academies and organizations as well as so many tens of thousands of scientists and researchers are ALL in a massive conspiracy.

        I will leave it up to you to decide if it’s possible that the warming set off an increase in CO2 that then “to use a terrible mixed metaphor” “Snowballed 😉

        Or if it’s more likely that 97% of the world’s scientists are all in a secret conspiracy

        1. Are you suggesting that climate change alarmists don’t pay their staff? Also, a “climate change denier” is almost by definition an imbecile, as it’s as unlikely for the climate not to change as it is for time to stop. More nuance is needed in decribing people’s stances. Some even go as far as to call them climate deniers, which only shows their own lack of intelligence. As for the CO2, there is definitely evidence of CO2 increases trailing as well as preceding higher temperatures. Higher temperatures generally cause conversion from liquid to gas so perhaps this has something to do with it. If you look back at data from millions of years ago, you can also see ice ages that correlate to increases in CO2 and it is evident that throughout most of Earth’s history the level of CO2 in the atmosphere has been several orders of magnitude higher than today and that the level today is close to extinction-grade low levels. Perhaps humanity’s contribution to increasing CO2 levels is actually a natural safety mechanism preventing the planet from reaching extinction-level lows.

    3. “Could years of data have been lost to melting or sublimation?”

      This is also my core question. I am wondering if you ever were answered satisfactorily? Certainly there could/should have been long periods of warm temperatures where many years of ice layers melted/evaporated. How can this possibly be known/accommodated?

      I should think this would have been one of the very first questions that came up when developing this measurement technique, but have not yet been able to find a meaningful response to this question. Please email me at if you happen to now know the answer. Thanks!

  2. Jim,

    I’m an ice core scientist currently working on my PhD, and have helped Bethan a bit with the content on this page. I’d be happy to try and answer these questions.

    For the first question, I’d say one of the coolest things about ice cores is that the records they contain are continuous from the present back many thousands of years. The oldest continuous record to date actually extends to 800,000 years before present (EPICA Dome C collected by a European consortium). For the gases contained in air bubbles in the ice, that continuity to today is somewhat complicated by the fact that bubbles don’t completely close and become isolated from the atmosphere until they are buried by 50-100 meters of snow and ice. The process of snow being compressed into more dense “firn” and eventually into ice takes anywhere from a few decades in places where snow accumulates quickly (lots of snowfall) to a few thousand years in places with very little snowfall. That means there is a bit of uncertainty around the exact age of the samples of atmospheric gas contained in the bubbles, as the upper few tens of meters of snow and firn is essentially open to the atmosphere. It also means that the ice surrounding the bubbles is older than the air inside. This is a primary area of research in ice core science.

    To connect the gas levels measured in those closed air bubbles with modern measurements, scientists pump air out of the porous snow and firn at regular intervals from the surface to where the bubbles “close-off”. This really does allow close matching between modern measurements and old air trapped in the firn and ice below. An Australian group did this quite nicely at a place called Law Dome, on the coast of East Antarctica.

    You are right to be careful assuming that the carbon dioxide, methane, or other gases inside the bubbles might not be perfectly preserved. It turns out that they are quite well preserved, especially in Antarctica. These bubbles in ice are the ONLY way that actual samples of the ancient atmosphere are preserved. For instance, since the Greenland Ice Sheet is in the Northern Hemisphere with most of the exposed land on Earth, the ice there contains high amounts of dust. Minerals in that dust do interact with gases preserved in the icy air bubbles, so much so that carbon dioxide records from Greenland ice cores are very difficult to develop. Antarctica preserves much cleaner, clearer gas records because it is very isolated from any of the few Southern Hemisphere land masses (and thus isolated from dust sources).

    I don’t think I’ll end up getting all of your questions figured out, but hopefully I’m sharing some helpful tidbits of the myriad details we work through in developing ice core records of climate and atmospheric composition.

    Ice core sites are picked very carefully to avoid too much complex ice flow at great depths, but there is always some. It is quite common that the bottom few hundred meters of deep ice cores (2500-3500 m long cores) are not included in climate records.

    Another relevant process occurs at depths greater than about 1000 m. The air bubbles in ice are actually compressed into the crystal structure of the ice, forming “clathrates.” This is an important process that also might affect gases in the ice, and so is studied very carefully. I don’t really know too many details of this, but the information is out there!

    Other things you mention are worth considering, and are considered by us icy scientists often. You’re right that snow does not constantly fall, so temperature records developed from stable isotopes of water (ice) may be biased towards the particular season of heaviest snowfall. We can study this during the modern period, where we’ve had satellites monitoring the weather/climate around ice core sites and hopefully get a handle on these biases. And, yes, especially in East Antarctica where snowfall is very low (less than 5 cm per year), wind scouring and sublimation can remove snow from the surface, resulting in (probably) short discontinuities in the data.

    Overall, I’d say that the reporting on all these difficulties is quite thorough in “the literature.” I recognise that’s not helpful to folks who don’t have access to scientific journals, but there is quite a lot of open-access data out there.

    In, fact, data from most of these ice cores are archived in a few places. My favorite is a website the US NOAA “Ice Core Gateway.” It has pretty much everything, so you can see yourself how many data points make up a 450,000 year dataset and that sort of thing (no, the gases aren’t sampled for every year going back that far).

    Hopefully that helps, and wasn’t TOO much information.



    1. I remember when the Artic not Antarctic was being researched thru ice core samples. Its strange because i was under the impression that the precipitation engine on earth wasn’t cranked until approximately 4800 years ago (meaning no precipitation any where on the planet
      How do we know there wasn’t another time in recent history were a like melt down of the Artics didnt occur. I think you human penguin guess at alot of data that you research. Without precipitation how would or could you even guess at what a year would be. Oh well theres a sucker born every minute i suspect

      1. Bethan Davies

        Dear Lee,

        Many thanks for your comment.
        Precipitation was occurring throughout the last ice age and certainly throughout the Holocene (last 11,000 years). You cannot build an ice sheet without precipitation! It was drier in places, such as around the margins of the ice sheets, but there was still precipitation. Although archives of precipitation are generally more complex than temperature, thicknesses of annual layers in ice cores provide some good information on precipitation. In places like the UK, lake and peat bog cores include microfossils which also provide information on past precipitation.

        I don’t understand your comment about a ‘melt down in the Arctics’. Please clarify. In Antarctica, there is a growing body of evidence that suggests the West Antarctic Ice Sheet has collapsed in the past.

        Please try to keep comments respectful and polite.

        Best wishes,

        1. Please forgive me if this was answered in the article. I’ve only had time to quick-read it at this time, bookmarked for later. Earlier I read another article on ice cores going back 800,000 years . Given your comments about evidence of ice fields collapsing, hope do you rule out total melt-off of an ice sheet? Would that not extend the time frame of the 800,000 bands, and add millennia to the core?

          Secondly, the question was raised about the necessity of precipitation to create the bands. Is it not a stretch to think that an 800,000 year core might have a significant number of years without precipitation?


      2. lynne morgan

        If the whole icesheet melted and you began again due to some (as now unrecorded) melting event, then the icecores wouldnt show mass volcanic events stretching back throughout 100’s of 10000’s of years. Which they do, and which can be correlated to other geological records such as sediments.

    2. shane brady

      G”day Peter, ime trying to find the ice core temp/CO2 graph[An Epica one i believe] that shows the entire last 800,000 years of fairly steady, flat Temp/CO2 lines[small ups and downs] with it rising markedly since the industrial revolution started burning all the FFs about 200 years ago.I want to add it to my data ime drilling him with. Ive seen it in the past, its one of the easiest to understand for most people.Ive studied a lot of data, but most find it very hard going.Most graphs need a decent understanding of the science that not many people have. You have to keep things as simple as possible for most people. Ime arguing with my brother about manmade climate change. Hes no dummy[nor am i]. Hes very high up in the fire brigade here in Tasmania, but i cant make him see how illogical it is to believe that 200 years of rampant FF burning and CO2 emissions have had no effect on the climate:/ Hard to discuss this with him as he gets very heated about the ” man caused climate change” …”beatup”..As he puts it:/ Bit of a downer that even an intelligent guy like him believes its all garbage, which it certainly is not, quite obviously:( I shudder to think whats going to happen when they get around to mining the methane hydrate and burning that!! Its coming!! Any idea where i can find that graph??

      1. Shane,

        The population of Europe 2000 years ago was estimated to be about 160 million. They were burning massive amounts of fossil fuels back then. Our current uses of fossil fuels are about 10x more efficient than theirs. Why is it only the past 50 years of fossil fuels that are a problem?
        Another point to consider is that every doubling of CO2 in the atmosphere is supposed to add the same amount of heat to our atmosphere. What this means is that for all of the warming we’ve supposedly caused with all of our fossil fuels, in order to induce the an additional amount of warming equal to what we’ve already induced, we’d now need to burn twice as much fossil fuels as we’ve burned so far. So we’ve got 1 degree of warming so far. We have to now burn twice as much fossil fuel as we’ve burned since 1950 in order to get another degree. Then we need to burn 4 times as much for an additional 1 degree.
        Maybe the truth lies somewhere between your take and your brother’s.

        1. John Ellerman

          Professor Will Happer is an expert on CO2, as a researcher on CO2 lasers. He says that the narrow wavelengths of light that cause the Carbon atom in CO2 to resonate either up and down or side to side, thereby generating heating, have almost all been absorbed already and a physicist has calculated that a doubling of CO2 will increase the temperature of the earth by a mere 1/50th of a degree, which is immeasurably small. Professor Carl Otto Weiss, advisor to the German Climate Board, has conducted a Fourier Analysis on all the European temperature data from 1850 to the present time and also the Vostok ice core proxy temperatures going back hundreds of thousands of years. He found that 6 recurring cycles, but predominately 2 cycles (one 28 year and one 230 year cycle) account for all the anual temperature fluctuations. Since CO2 has basically trended continuously downward it does not figure in the data as a driver of temperature. The level of CO2 in the atmosphere has increased from 0.025% per-industrial revolution to 0.04% now. This is a minuscule amount of a weak greenhouse gas and is perhaps a fifth or tenth of what it was when plants evolved. If the CO2 had continued to be removed by the Foraminifera then all plant life would have become extinct because they can’t survive below 0.017%. Is it not reasonable to suggest that our CO2 production is actually saving the planet?

          1. I agree that there’s an argument that humanity’s contribution to increasing CO2 levels may be helping to save the planet. After all, we are as much a part of nature as anything else – a fact that seems to be denied, ignored or forgotten by many people. Although the dictionary separates humanity from nature, this is a concept invented by man, likely through religious beliefs. There is in fact no real evidence or logic to conclude this – where is the separation between the inanimate and the animate in evolution, between the organic and inorganic, the non-sentient and the sentient, and who is to say that the tools built by humans are any less natural than those used by a bird to build its nest, or a termite to build its mound. Maybe Humanity is nature’s way of correcting the low CO2 levels.

        2. Seems a bit of a stretch that 160 million people burning fires at night emitted more carbon than 7.7 billion people burning coal 24/7.

    3. I have a question how does it work and what are the names or the tools and objects you use

    4. Patricia merritt

      Why don’t scientists wear gloves and masks when handling deep sea chores..aren’t they afraid of unknown bacteria causing diseases ?.

      1. Bethan Davies

        Generally, gloves and googles are worn when opening the cores, to avoid contaminating the insides, but this PPE should also provide protection for the scientist.

    5. A Harvard University study (included several other institutions recently concluded that air pollution had caused 7-8 million deaths. Seems like there is more reason than just climate change to not pour accumulating greenhouse gases into the atmosphere. NASA reported that pollution reduced by 30% over New England when we were under lockdown. Other locations on the planet remarked similarly.

      1. Totally agree. Reducing pollution should be the main driver. It is something that is actually tangible and will rally far more support than threats by climate alarmists lobbying to destroy our way of life, which is what stopping all fossil fuel burning amounts to. The focus should be on reducing pollution by emissions that are actually harmful to life, rather than an abstract assumption that the changing climate is bad for the planet

    6. David Warrilow

      I hope this reaches you after so many years. You said:
      >The process of snow being compressed into more dense “firn” and eventually into ice takes anywhere from a few decades in places where snow accumulates quickly (lots of snowfall) to a few thousand years in places with very little snowfall.
      The critical question is how long does it take for the Vostok Ice Cores to seal? I have found various values, with one being up to 300 years. I realize that the mixing is complicated but if an ice core sample would be taken a thousand years from now, what would it say the current CO2 level be? Wouldn’t it be some value that is a 300 year average?

  3. Hi I am doing environment studies for building at Unitec Mt Albert Auckland NZ, just would like to know if the ice core samples that have been recorded are from the same altitude around the planet. The atmospheres would change the density of the samples and would like to know what the out come of the samples would return at different altitudes.

    1. Dave Pene, I’m not an expert but since nobody else responded I’ll try to answer that. Ice cores are generally taken where the ice is thickest, meaning the top is the highest around and the bottom is the lowest they can find. Here the ice flow is slowest and the ice is ideally the oldest. That means it will be about two miles high in Antarctica, and over a mile high in Greenland. So, the air will definitely be denser at Greenland, but that doesn’t make much difference except that there might be more or larger bubbles there. What the researches are interested in is the content of the bubbles–not their size or quantity. They are assumed to retain the relative gas ratios of the atmosphere that they trapped, as well as the isotope ratios. These ratios don’t vary much in the troposphere–the gasses are pretty well mixed. –AGF

  4. In the legend on figure 1 (copied below) — is says the current period is at left — shouldn’t this read — the current period is at right ????? The current time — (i.e. 2014) is at the right of the graph right ?

    420,000 years of ice core data from Vostok, Antarctica research station. Current period is at left. From bottom to top: * Solar variation at 65°N due to en:Milankovitch cycles (connected to 18O). * 18O isotope of oxygen. * Levels of methane (CH4). * Relative temperature. * Levels of carbon dioxide (CO2). From top to bottom: * Levels of carbon dioxide (CO2). * Relative temperature. * Levels of methane (CH4). * 18O isotope of oxygen. * Solar variation at 65°N due to en:Milankovitch cycles (connected to 18O). Wikimedia Commons.

    1. Bethan Davies


      The isotopic record of past climate recorded by the bottom layer of the ice core is determined by the age of the ice. The age of the bottom layer of the ice core is determined by the depth to which the ice-core is drilled, and the thickness that one annual layer represents.

  5. Notably missing is a mention of the temporal relationship between CO2 concentrations and the derived temperature. Cross correlating the two from the Vostok data set reveals a quite variable delay between local minimums and maximums of CO2 and temperature, with CO2 lagging by up to 800 years. The DomeC data is more precise and definitive about the delay, which is on the order of a couple of hundred years, although slightly asymmetric. The delay can only be indicative of biology catching up with a more favorable climate as it takes time for CO2 to accumulate up to a large enough level to support a larger biomass. Similarly, on the down slope, the extra CO2 temporarily sustains a more robust biomass as the planet cools.

    Also missing is any mention of the strong correlation between the Earths variable precession, orbit and axis and the temperatures extracted from ice cores, especially DomeC whose temporal positioning of ancient samples is far more accurate than Vostok.

    1. Mark Thogerson

      The lag time between can be explained by ocean/atmosphere interactions. CO2 is very soluble in water, but its solublility decreases inversely with temperature. Also, as the earth warms, most of that energy is absorbed by the oceans. Early in a warming cycle, atmospheric CO2 increases modestly, because much of it dissolves in the ocean. As the ocean warms (more slowly than the atmosphere due to its high specific heat) it stops absorbing the CO2, and atmospheric CO2 increases more rapidly. Warming phases are initiated primarily by “positive interference” between Milankovitch cycles, with atmospheric CO2 providing positive feedback at some point. If current conditions were due entirely to Milankovitch cycles, global climate should be cooling slowly for another few thousand years, just as it has for the past 9000 or so.

    2. Radoslav Porizek

      It was on this page before, then it was wiped out.

      The google search is still providing this page with the following text:
      “Antarctic Glaciers › glaciers-and-climate
      Mudelsee (2001) – Over the full 420,000 year Vostok history Co2 variations lag temperature by 1,300 years ± 1000. Caillon et al 2003 analysed the Vostok …”

      I have serached for: co2 is lagged 300 years behind temperature

  6. Here are some informative plots. They show a metric congruent with the probability that given a change in some variable (for example, temperature) that in N years (0-10K in these plots) the same variable (autocorrelation) or a different variable (cross correlate) will be changing in the same direction (positive values) or in the opposite direction (negative values). A value of zero means that there is an equal probability that the variable will be changing in either direction. Results from both Vostok and DomeC shown. The peak in the green is at the delay where changes in CO2 are most correlated to changes in temperature.

    The green line consistently above the magenta line shows that changes in CO2 concentrations are always more correlated to past temperature than future temperatures. Note that for the DomeC data, with better temporal positioning, future temperatures are nearly completely uncorrelated to past CO2 levels. This next plot adds CH4, showing that it is delayed by even more and is another unambiguous biological marker.

    These autocorrelation tests of temperature data show the strong correlation with various periodic orbital attributes.

    correlation to the 41K year period of axial tilt variability

    correlation to the 110K year period of variable eccentricity

    Correlation to the precession of perihelion is shown here (around 20K years) although also present is the strong correlation to the period of axial tilt variability.

    This last one shows the temperature variability plotted along with the axial tilt and variable eccentricity.

    The smoothing applies is averaging around center and is used to 1) normalize the sample period between recent samples and ancient samples, 2) normalize the sample period between variables with different temporal resolution and 3) to act as a low pass filter to remove short term correlations to reveal longer period correlations.

    I should have provided enough information to replicate these results, but if more is required, more is available.

    1. Hey, this is a great question highlighting one of the tricky details in this work.

      The plot you linked to shows delta-Deuterium (dD) in blue, which is the ratio of H2 to H3 (called deuterium) in H2O (water).

      The orange curve, however plots delta-18-Oxygen (d18O, ratio of 16O to 18O) ratio of atmospheric O2, NOT the delta-18-Oxygen ratio of H2O. This is a very important distinction.

      So the orange curve represents the isotopic composition of atmospheric oxygen measured in air bubbles trapped in the ice cores, versus the blue curve of the isotopic composition of water (a.k.a. ice). This proxy is representative of global ice volume, because the size of the ice sheets determines the d18O (and dD) composition of seawater, which in turn sets the isotopic baseline of the global hydrologic cycle.

      This baseline isotopic signature translates to atmospheric oxygen through photosynthesis—even for plants, “you are what you eat.” Basically plants respire O2 that reflects the isotopic signature of the H2O they depend on for life! Check out what’s called the “Dole Effect”

      If the blue line was indeed dD of H2O, it would be very similar to d18O of H2O in the Vostok ice core. There are very small, useful differences in how O and H fractionate in water which can tell us a bit about where the moisture that falls as snow on the ice sheets comes from. This is called “deuterium excess.” Some details are here:

      Hope that’s helpful!


      1. To be absolutely clear, when I say, “This proxy is representative of global ice volume…” I am referring to d18O of atmospheric oxygen, which is the orange curve on the plot Alex linked to.

        1. Whew, looks like I posted too fast! In the last paragraph I meant to say that if the ORANGE line was indeed d18O of H2O, it would be very similar to dD of H2O at Vostok (blue line in Alex’s plot). Apologies for my confused response.

  7. Question for those knowledgeable about the actual data. NASA has an interesting statement on their climate change website that says “For 650,000 years atmospheric carbon dioxide had never been above this line [roughly 300ppm]” (source: I looked at a handful of datasets on the NOAA ice core website which are for periods of over 150k years. They generally have a resolution in the hundreds of years (with some exceptions of higher resolution). When looking at a trend over hundreds of thousands of years, plotting data points every couple hundred years makes plenty of sense. But I am curious about whether there is higher resolution data for the specific historical periods during which there are rapid increases in CO2 levels, specifically -130k, -250k, and -330k years. I wonder if the data sampling frequency is causing aliasing which doesn’t allow us to see the true peaks that have occurred in the past. It seems strange to me to draw such strong conclusions like NASA’s above when comparing data from the past 60 years, which are on an annual timescale and which also coincide with the periodic rising trend, with those of the past. I’m curious whether this has been studied or acknowledged by the community performing this research. It’s very fascinating and impressive work, so thank you to all the dedicated scientists and engineers who are working in the field!!!

    1. I had the same question. Can we get some adult supervision to explain NASA’s conclusion in terms of temporal resolution? Also, to what degree has each ice core data set been cross-referenced to each corresponding data set by location and researcher (ie do we have a list by year of all of the available ice core data for that year to achieve a reasonable average?) If not, which is considered definitive? Lastly, how well does ice core data from different locations correlate?

  8. “You are right to be careful assuming that the carbon dioxide, methane, or other gases inside the bubbles might not be perfectly preserved. It turns out that they are quite well preserved, especially in Antarctica.”

    So you evidence is the phrase “it turns out”. Ok. Well…It turns out the air bubbles are NOT well preserved, especially in Antarctica. See I know how to use the phrase “it turns out” just as well as you.

  9. Alice Freund

    I am studying to be a science teacher and have been assigned to ask an expert about the project I am doing for class. The project is making a poster about ice core sampling. Is there anyone on this site who could refer me to an expert or is an expert? If so I would like your/the contact info and I will be asking for a CV. That would be quite helpful. Thanks.

  10. Alice Freund

    Specifically I would like to ask an expert 1) is there ice core data that shows the very recent exponential rise of CO2 to over 400 ppm in the last few years, and 2) about the controversy of why the CO2 levels have historically appeared to follow the temperature.I have to cite the expert so I will need more info about your expertise (see above). Thanks!!

    1. Alice, great questions!

      For the first, some of the best ice core CO2 data that overlaps with modern measurements comes from a place called Law Dome in East Antarctica. Dr. David Etheridge from Australia is the expert on this, and there is a very detailed article about his work here:

      As for the second question, natural fluctuations of temperature in the past have indeed led CO2. There is a whole field of research trying to understand “climate sensitivity,” which is simply asking how much temperature increase can be expected per increase of CO2. For instance, the most recent Intergovernmental Panel on Climate Change (IPCC) assessment report nicknamed “AR5” puts climate sensitivity likely between 1.5 and 4.5 degrees C of warming per doubling of CO2.

      While it is somewhat of an overwhelming document, the entire IPCC AR5 summarizing the state of the research is available for FREE online.

      Rather than citing a conversation in the comments here, even if I am an expert, I would encourage you to cite credible information you can find on these topics. The links I have provided are for very reliable sources which can confidently be cited.

      Hope that helps!

  11. How do scientists get this information?

    Where can scientist collect data from ice cores ?

    What can ice cores tell me about past climates ?

    What is an ice core?

      1. Approximately how many years of time is covered by each CO2 reading from an ice core? Comparing CO2 measurements which may represent the average conc. over a 1000 year period of time, with daily CO2 measurements could be very misleading. You would miss any large spikes or dips that may occur due to natural forces.

        1. Bethan Davies

          In the upper parts of the ice core (last few hundred years), annual laminations in the ice allow us to derive annual CO2 and isotopic variations. As the ice is compressed deeper in the core, the annual layers are lost so several (not 1000s) of years may be amalgamated.

          1. It would be useful to know how big ‘several’ is. Specifically, in the EPICA core, how many years make up a data point at 200kyr, 400kyr and 740kyr ago.

  12. John Anderson

    I’m not a scientist and never will be! Jeremy Corbyn’s brother Piers was given a slot on TV last week and seems to maintain that Global Warming is no longer taking place and CO2 is not the ‘driver’, but sunspots (I paraphrase!). Is there an easy answer to this “heresy”?

    1. I would point you to the Technical Summary of the UN Intergovernmental Panel on Climate Change 5th Assessment Report. The entirety of this most recent AR is freely available from, and the Working Group 1 “Physical Science Basis” portion is where you will find answers to these scientific questions.

      Take a look at page 54 of the Technical Summary, here: Although the language and material in this summary is still at a relatively high level, you can still gain a lot by just looking at the figures.

      Figure TS.6 (also TS.7) displays the “Radiative Forcing” (RF) of climate change during the industrial era, from 1750-2011 (the report was published in 2013). IPCC defines radiative forcing as “a measure of the net change in the energy balance of the Earth system in response to some external perturbation. It is expressed in watts per square metre (W / m-2).” These figures essentially show the energy budget of the Earth, keeping track of warming and cooling just like you’d keep track of cash flow into and out of your bank account.

      We can see that the change in RF from CO2 is +1.5 to 1.86 W / m-2 over the industrial period. Compare that to changes in “solar irradiance” which includes sunspots: less than +0.1 W / m-2. Solar variability has at least a 15x smaller effect on RF than CO2! Greenhouse gases are really the primary drivers of warming by a significant margin, importantly with relatively small uncertainty

      There is an 11-year cycle in sunspot numbers, which has been observed for several hundred years (you can see these data in the top box of figure TS.5). We can’t very well anticipate future solar variability, but I quote from p.56: “there is a high confidence that 21st century solar forcing will be much smaller than the projected increased forcing due to well-mixed greenhouse gases (WMGHGs).”

      Hope that helps in some small way. I’m not explaining all of the underlying fundamentals (i.e. physics) of why GHGs have such a large RF compared to solar irradiance, but if you are craving that just dig into the full Physical Science Basis report! (

      There is a lot of technical information in the IPCC reports, and it can sometimes be a bit overwhelming, but it represents fairly well the entirety of our understanding of the Earth system as compiled by the expert authors and editors as well as thorough peer review.

  13. The ice-cores give a recording of variations of temperatures at the site, but only at the site of the cores. (CO2 would be the same around the globe.)

    How do those temperature variations relate to those at other parts of the globe? For example, how does a Vostok ice-core record of a change of say 9C translate to other latitudes?

    1. This is a very insightful question. You are correct that CO2 would be the same around the globe, as it is a well-mixed greenhouse gas (GHG). We see that trace gas records, particularly CO2 and methane (CH4), the two most important GHGs in terms of radiative forcing (see my response to John’s question above), are essentially identical across Greenland and Antarctic ice cores. Methane records in particular are used to synchronize deep ice core records.

      The easiest way to explore this question of how Antarctic temperatures relate to the rest of the world is to look at what Greenland ice core temperature records look like. Here is a great article on explaining state-of-the-art data from the WAIS Divide ice core from West Antarctica (I played a small part in this study and am one of many co-authors):

      This study explores what is called the “bi-polar seesaw,” a fundamental aspect of which is the question titling the article: “How long does it take Antarctica to notice the Northern Hemisphere is warming?” It turns out that temperature changes in the northern versus southern hemispheres are actually out of phase due to how long it takes one or the other hemisphere to respond to a temperature change in the other.

      While you weren’t asking about this, the article does plot temperature proxy data (delta-18O, a fundamental measurement of the isotopic composition of the ice). You can think of delta-18O basically as temperature, but calculating exact degrees-celsius from this proxy is more involved than I am able to go into! Briefly it involves measuring the physical temperature of the ice sheet using the borehole created by the ice core, which then helps better get an estimate of true temperature change through some interesting math and physics.

      Anyway! You can see in the first figure in the article than Greenland temperature fluctuates much more quickly and with greater amplitudes than Antarctica. This has to do with the geography of these places, and energy (heat) transport through slow ocean circulation processes. Consider that Antarctica is a large continental ice sheet surrounded by the Southern Ocean and from this alone it makes sense that temperature changes will be more gradual in the Antarctic. It takes a long time to get an entire ocean to warm or cool, or for ocean currents to transport water warmed in the North Atlantic all the way to Antarctica–around 200 years, based on this new data from WAIS Divide.

      That really only partly answered your question, but I wanted to be able to point you to freely available data so you can see the difference between Antarctic and Greenland ice core temperature proxy data. We also know that polar amplification causes more rapid warming at high-latitudes due to dominant pole-ward circulation and thus heat transport. See discussion in IPCC 5th Assessment Report WG1 Chapter 5, Box 5.1 on page 396, here

  14. So from the Volume change data above (Ice Age Temperature Change figure), I would estimate that the earth will possibly start a new ice age in 1K to 5K years from now. We have been in a melting phase for about the last 15K to 20K years. This is cyclic. So, given the talk about methane, CO2, H2O (yes water is a green house gas too), sun activity, magnetic field of earth, volcanic activity under Greenland and the Antarctic, cows farting and adding to the methane part of green house gases (yes a farming factor), plus industry and the growth of cities, plant growth (when CO2 builds up in the atmosphere, plants actually thrive), the earth wobbling influence, etc…. I am focused on the cyclic nature of the ice volume.

    Would someone that works with the ice core data (maybe from the OSU as I know OSU has faculty working on ice core analysis)…. what are your thoughts about another ice age coming as we are nearing an ice volume local minimum and may at some time (1K to 5K ish years) see the earth ice volume begin to go in the “high” direction as may be extrapolated from the graph.

    Thx for your time.

    1. Hi there, although I myself am not at OSU, I can provide some perspective from a faculty member there (and his co-authors). Earlier this year Peter Clark and others published an interesting article in Nature Climate Change which provides long-term perspective on, climatically, where we’ve come from over the past 20,000 years (from ice cores and other archives) and where we may be headed in the next 10,000 years depending on 21st century policy decisions.

      They argue that this longer-term framing of past and future climate change better informs decision-makers and the public that anthropogenic global warming is not just going to be a problem on the timescale of the 20th and 21st centuries. We usually only look at 150 years of historical temperature measurements to establish that the climate is warming, for instance. But thanks to ice cores and other archives we know what temperatures have been for several hundred THOUSAND years, and we also know that the input of greenhouse gases (GHG) into the atmosphere now will have consequences (in the form of higher temperatures and sea levels) for the next 10,000 years and beyond*.

      The article itself is here (unfortunately probably with a pay-wall):

      See a Washington Post write-up here:

      Between our emissions of CO2 and also, as you say, methane from cows farting (FYI, I believe it’s ruminant belches that are actually worse), we are stepping outside the normal orbital-driven ice age cycles (called Milankovitch cycles; excellent overview from Skeptical Science: Sun activity, magnetic field, volcanic activity are lesser terms.

      I guess I share this in hopes to provide more than just an answer to your question.

      The answer is that NO we are not going to have another ice age, and this article takes it further to really demonstrate what we are going to face instead. Similar to my answer to John Anderson’s question above, the total radiative forcing from Milankovitch cycles, which primarily caused the ice ages, is much less than the forcing from CO2 (see the Skeptical Science article for more). We’ve very quickly, over the last 200 years, stepped right up into a new normal where we have 400 ppm of CO2 in the atmosphere instead of the usual ~280 ppm during warm (interglacial) periods—we know this from Antarctic ice cores, as you saw in the first figure Bethan presented in this article.

      Based on this study (which agrees with many others), depending on what we decide to do about GHG emissions in the next decade or two, we are choosing between having anywhere from +1C to +6C warmer global temperatures for the next 10,000 YEARS. Not going to grow any ice sheets in those conditions (in fact we risk completely losing all of Greenland’s ice and most of Antarctica’s!).

      Based on policy decisions about GHG emissions in the coming decades, we are committing ourselves to possibly having anywhere from 10 m to 20 m of sea level rise over the next 10,000 years. These are large and very long-lasting decisions we have to make NOW—namely to stop emitting warming greenhouse gases ASAP—and we have to realize that if we decide to let ‘er rip with GHG now that we will negatively impact future generations and change the face of the earth for many millennia to come.

      *Once you put CO2 into the atmosphere, it stays there for 500 to 1000 years because trees don’t uptake CO2 on long timescales (they eventually die and return the CO2 to the atmosphere) and the other major CO2 reservoir, the ocean, becomes too full of CO2 (saturated) at its surface and can’t quickly remove CO2 from the atmosphere (hence 500 to 1000 years to get CO2 into the deeper ocean).

  15. Michael O'Brien

    I worked for many years in a National Association of Testing Laboratories (NATA) registered lab in Australia. We were required to quote UncertInty of Measurement within a stated Confidence Interval for our tests. I would like to see this information for data on carbon dioxide and other gases in ice cores, particularly for oxygen isotopes at extremely low concentrations. I note that this information dates back many years. Surely more modern measurements would have lower uncertainty.
    Also what about the uncertainty of temperature measurement in past centuries.

    1. Michael,

      It’s great that you bring your experience in laboratory testing to considerations of uncertainty in ice core data.

      There are a number of factors affecting the precision and accuracy of ice core measurements, which are very carefully documented and presented in the literature.

      In many cases, with the progression of technology, the biggest limiting factors are no longer in the instruments used (i.e. mass spectrometers, gas chromatographs, cavity ring-down spectroscopes). That statement applies to the more routine measurements made, including CO2 concentration of ancient air trapped in bubbles in the ice, and oxygen isotopes in the ice itself which provides a temperature proxy. More advanced techniques, for instance breaking down the carbon isotopic composition of that CO2 to name just one, still have relatively large analytical uncertainties.

      Other uncertainties in ice cores arise through a number of factors.

      One is the development of ice core timescales, which are a combination of annual layer counts, absolute dating of volcanic horizons, ice-flow models, and gas chronology matching.

      There is also uncertainty in diffusion of chemical signals in the snowpack, which essentially averages these signals on depth scales controlled by site temperature and snow accumulation rate. Diffusion is studied and documented so scientists know the minimum resolution at which they can interpret actual climatic or environmental signals rather than meaningless noise.

      There is also uncertainty of the spatial coherence of chemical signals in the snow (i.e. blowing snow, snow drifting which forms small dunes called ‘sastrugi’), variation in the timing of snowfall and so on.

      Some of these sources of uncertainty are briefly discussed here by Eric Steig of the University of Washington (full disclosure, Eric was my M.Sc. supervisor).

      For an example of the state-of-the-art of ice core dating including uncertainty, check out two papers on the WAIS Divide ice core timescale. This is the highest-resolution Antarctic ice core record spanning the last 68,000 years. These were both published in the European Geophysical Union journal ‘Climate of the Past,’ which is open access and also features open paper discussion and peer review.

      The WAIS Divide deep ice core WD2014 chronology – Part 1: Methane synchronization (68–31 ka BP) and the gas age–ice age difference

      The WAIS Divide deep ice core WD2014 chronology – Part 2: Annual-layer counting (0–31 ka BP)

      This is a lot of information to digest, but I will emphasize that these sorts of detailed presentations of uncertainty associated with ice cores are published for all major ice core projects. All scientific journals require such presentation of uncertainties associated with all presented data. Additionally, as an international community, the International Partnerships in Ice Core Sciences works to maintain high standards for presenting uncertainties affecting these valuable data.

      Hope that helps a bit!



  16. Fred Yeganeh

    Is the current rise in global temperatures statistically significantly greater than the natural variation in Greenland ice core temperature variation seen over the last 10,000 years? From graphs that I have seen, the current rise in global temperatures is well within normal variation where as the CO2 rise is obviously a dramatic new change. If this dramatic rise in CO2 has not caused any statistically significant abnormal rise in temperature when compared to a 10,000 year record, it is unclear how much an effect this rise in CO2 is having. I guess the only answer is that the rise in temperature is lagging the rise in CO2. I would like to understand how the Greenland data show no abnormal significant rise in current temps. Thanks for any insights!

    1. Fred, I suspect you may have seen mislabeled Greenland ice-core graphs based on Alley 2000 and the Cuffey and Clow 1997 papers. They are all over the internet, typically found on blogs that try to refute anthropogenic global warming. The last data point from these ice-core studies was 1855, 161 years ago, however the graphs are mislabeled labeled as current or present.

    2. Hi Fred,

      Your framing of this question is great, very clearly set out. You are right that we have dramatically increased CO2 concentration in the atmosphere, from about 280 ppm before the early 1800s to about 400 ppm today (which we know from, guess what….ice cores!).

      We absolutely expect a temperature rise from this CO2 (and other greenhouse gases) we’ve put in the atmosphere, based on physics that Svante Arrhenius provided us with back in 1896. There is indeed a lag in the temperature response for the giant cruise-ship that is Earth (it’s not a nimble speedboat, quick to change course). The massive oceans take up much (>90%) of the additional heat trapped by increased CO2 in the atmosphere, among other causes of what seems like a slow atmospheric temperature rise.

      There were some interesting long-timescale (~1000 year) natural climate variations over the last 10,000 years, including a “climatic optimum” 9,000 to 5,000 years ago, the Little Ice Age several hundred years ago. The climatic optimum is likely a continuation of the solar forcing that we call “Milankovitch Forcing” with a maximum of Northern Hemisphere heating at this time due to the high angle of Earth’s obliquity at that time. Since then we’ve had orbital forcing favoring cooler temperatures, but we’ll see what we can do about that with our greenhouse gases.

      There are also higher-frequency patterns of natural climate variability like El-Niño/La-Niña, to name only one, which factor into the natural fluctuations of global and regional climate. Despite this noise, we are indeed seeing anthropogenic warming emerge from natural variability.

      A very detailed demonstration of this, unsurprisingly, comes from the Intergovernmental Panel on Climate Change. This is from the 4th Assessment Report (although there is a version of the same figure in the newer 5th report, but I don’t think it is as clear).

      These two plots show the observed 20th century temperature trend in black lines, with the results of 14 simulations by Atmosphere-Ocean General Circulation Models (AOGCMs) run with anthropogenic and natural forcing (yellow lines in the top plot) versus the same AOGCMs run with ONLY natural forcing (blue lines in the bottom plot).

      You can see from the blue AOGCM results, that you cannot produce the 20th century warming trend without including anthropogenic forcing, which includes greenhouse gases and also aerosols and other pollutants—which actually have a cooling effect. If you don’t include anthropogenic aerosol cooling, the models over predict the observed warming. We understand very well what is going on! There is still some rather unpredictable natural variability, but the budget-keeping adds up.

      In the case of Greenland temperatures specifically, we are seeing the trend begin to emerge out of natural variability. Events like the summer 2012 melt event which spanned the entire Greenland Ice Sheet are rare but not unprecedented—a similar event occurred in the 19th century. However, they are very likely to become more common in the near future as global temperatures increase, sliding the bell curve of temperature variation towards the hot end of the scale. In the polar regions where natural variability is particularly extreme, the emergence of clear anthropogenic warming is slower to emerge but in recent years is exceeding previous variability. The global forecast isn’t for anything but more heat…

      You can keep track of what is currently happening in Greenland here:, with lots of links therein to data on all things icy from the National Snow and Ice Data Center.

      Hope that was useful!


  17. Bethan Davies

    Question from Ning Tu:

    First I would like to thank you for your detailed introduction on ice cores! In my recent research on a particular topic, one question has become a key issue, and I believe you will have the answer: in current ice-core research, has the oxidation of methane in the air bubbles (trapped in the ice cores) been considered? I mean, at such low temperatures, the oxidation rate must be extremely low (ref: However, we are also talking about extremely long time, i.e., in the order of hundreds of thousands of years. This is important as the oxidation of methane produces CO2 and alters and composition of these gases inside the ice core with time, i.e., the oxidation makes the otherwise static air composition a dynamic process.

    What input do you have? Thanks!


    1. Bethan Davies

      Reply from Peter Neff:


      I am fairly sure someone has considered methane oxidation in air bubbles…. but a brief search through the literature didn’t turn up much.

      The fact that methane records across many ice cores are nearly identical suggests there are not anomalies caused by oxidation or other such processes. Methane is a well-mixed gas in the atmosphere (with a lifetime of about 9-10 years), so both Greenland and Antarctic methane records are very similar. So similar, in fact, they are matched and used to improve dating of the ice core records (for example, Blunier and Brook, Science, 2001). Careful studies have also been performed to explore how the air bubbles in ice transform to clathrates deep in the ice sheet (bubbles are forced in to the crystal lattice due to extreme pressure), and how this affects gas measurements from the bubbles.

      We do also ensure that the ice cores, once extracted, are maintained below -20ºC to prevent any gas loss or other reactions possibly including oxidation.

      You mention that oxidation of methane would produce carbon dioxide, affecting this important record from the ice cores. Again, we don’t see this across the Antarctic CO2 records, but we do know that we can’t get reliable CO2 records from the Greenland ice cores. This is because the high levels of dust in the ice, which is partly composed of calcium carbonate (CaCO3), cause additional production of CO2 in the ice itself. This is possibly also related to organic species reactions in the ice. Because the Antarctic is so isolated from dust (the Southern Hemisphere is ~10x less dusty than the Northern Hemisphere mostly due to the reduced land area), the ice is incredibly clean and we have well-preserved CO2 in the air bubbles.



    2. I’d add to that, what about simple chemical processes, particularly interaction with the ice itself with the gasses in bubbles. With pressure and time, does the ice act as a perfect solid? Is co2 exchanged with the gas bubble and the ice itself?

      Thanks in advance,


  18. I am currently in South Australia in the extratropical cyclone. WA also snowed in Bluff Knoll and NSW is going to get the wild storms as well.. Can anyone tell me if the drilling in the Antarctic is having an affect on our current weather in the southern hemisphere. Just a thought as I think the seasons are a month off as well and should have new dates.

  19. Robert Yoder

    I’ve read that the upper layers of the cores have visibly distinguishable layers, but as you get deeper, the layers become visibly indistinguishable. If that is true, how many many years of data are visually discernible?

    1. Bethan Davies

      It depends very much on the precipitation and flow regime. On the James Ross Island ice core, the first few hundred years are easily discernible.

      1. Bethan is right, it depends a lot on how much snow falls! Where you have tens of centimeters of snowfall each year you can generally see the annual layers if you know what to look for. We sometimes dig pits into the recent snow, and with back-lighting you can see layers–google “Antarctic snow pit” for some great examples. Based on how the winter versus summer snow “packs” the layers are distinguishable (windier in the winter means smaller broken bits of flakes pack more densely, I believe). Also, in Greenland ice cores the layers tend to be very easy to see because there is much more dust in the atmosphere in the Northern Hemisphere summer (when so much land area becomes snow-free). These dusty layers are darker. Once you get a few hundred meters deep, the ice becomes very glassy and more homogeneous to look at, in my experience, but under the right light conditions experts can still see the layers. Thankfully, we can much more easily “see” annual layers in the chemistry of the ice and can count back ~40,000 years.

  20. Wow, I hope that you all have not gone and shot yourself since Mr. Trump has been elected. But seriously, no posts since July has me wondering what happened.

    Like (I believe) most of us who don’t trust a politically motivated, politically funded science community, I have been a skeptic regarding how much mankind is contributing to climate change. I will read a lot more but some of you are convincing me that humans are likely causing things to change too rapidly.

    The problem is, what can we do about it? Force upcoming industrial giants in other countries to modernize? Convince developing countries to wait until we have a solution (free energy, etc.)? Kill cows? What would you do? It seems to me that we can’t (reasonably) do enough. This ends up being a politically driven tool to subdue and control American industry. America has already cut GHG a LOT folks. How will you get the rest of the world to do what is necessary? Frankly I believe your talents might be best used to help the world figure out how we might remove as much GHG as possible (if that will slow things down at least).

    Scientists in your field of study should all get together and educate the public about what exactly you DO know, what you theorize and what can be done to address it. This is not doing that, it needs to be on TV and promoted by all sides. Not just helping one political party destroy supporters of the other party (I don’t like, or support, either one, by the way) but honest debate and discussion on what we can actually do to affect climate change in humankind’s best interest.

    I know this is off-topic but we all know the “elephant” is in the room (pun intended) even when we’d like to pretend it is not sitting right behind the donkey.

    1. Bob,

      Luckily this website and ice core research are international in scope, so turning political tides in one country don’t (directly) affect us all.

      Ice cores provide clear context for how far off scale modern greenhouse gas concentrations are. It has been the work of many other climate scientists to demonstrate with high confidence that most of these greenhouse gases are emitted by human activity.

      I find this data presentation by Bloomberg (of all places) to be one of the best illustrations of just how big a role human greenhouse gas emissions have played in warming since the 1950s. It breaks down the individual climate drivers adding up to the observed upward global temperature trend.

      Unfortunately the article is framed as an argument against skeptics, which makes it somewhat adversarial, but this is the state of discourse in the United States.

      As far as action, you are correct that this requires significant action as quickly as possible. You bring up thoughts of equity between developed and developing nations, which plays a large role in international (UN) negotiations towards action. There is also the issue that certain countries are feeling and will feel the effects of climate change more quickly and/or more strongly than other countries (read: low-lying and island nations). This all has to be taken into account to reach agreements such as the Paris Climate Agreement.

      As an American scientist I have certainly been reflecting more on advocacy in recent months. My most simplest point: we must support international action towards reducing GHG emissions, and the Paris agreement begins to do this. It is not enforceable enough, nor strong enough, but it is the framework built thus far and must continue. We do need to get out in our communities and get this message across.

      Thanks for your comment and for considering the information presented!

  21. Typically how long do interglacial periods last? I’ve read that it is about 10,000 years. The last ice age ended about 11,000 years ago. Are we entering an ice age period or are we already in one?

  22. Typically how long do interglacial periods last? I’ve read that it is about 10,000 years. The last ice age ended about 11,000 years ago. Are we entering an ice age period or are we already in one?

    1. If you look at the plot of the Vostok record up at the top of the page, you can see that warm interglacial periods tend to last several 10’s of thousands of years (say 10,000 to 30,000 years). The reason for the glacial-interglacial cycles are slight changes in Earth’s orbit around the sun, which affect the amount of solar energy arriving at Earth’s surface at different times of the year. You can calculate this quite well, and if you look at the Wikipedia page for “Milankovitch cycles,” named after their discoverer, there is actually a graph of past and future orbital cycles (

      We are currently in an interglacial period, which we call the Holocene. It began ~10,000 years ago, and during this time modern civilization as we know it has emerged. Stable, warm climates are really helpful to grow a population and develop agriculture, languages, etc.!

      If we had not emitted so much carbon, we would be on our way back to an ice age in a few thousand years. However, we have increased atmospheric CO2 by about 40% since the Industrial Revolution of the 1800’s. The added heat trapped by these and other greenhouse gases will now combine with the natural changes in solar forcing from orbital changes. I don’t know the exact result of this new energy balance, but it is safe to say we can expect a significantly modified future. I believe global temperature change from an ice age to a warm period has tended to be about 3-5ºC. Since 1850, we have measured a global average increase in temperature that is almost 1ºC and rising. International efforts such as the Paris Climate Agreement are trying to limit human-caused warming to 2ºC. That might give you a sense of our impact–in just ~200 years–relative to the natural fluctuations we’ve seen in the past which occur over 1000s to 10,000s of years.

      Hope that is useful!



  23. Nicolai Ellingsen

    Hello and thanks for an informative web site. I found answers to a lot of questions here, but not the one I was looking for and I hope you can help me satisfy my curiosity.

    I see that the core samples contain melt and freeze layers, but how do we know that layers of ice has not melted completely and disappeared from record in warmer periods? If this was the case then the sample of historic greenhouse gas and temperatures would be censored, that is warm periods with a lot of carbon in the atmosphere would not show up in the core samples.


    1. Nicolai, this is an excellent question!

      One way we can tell whether or not there were out-of-the-ordinary melt events during past warm periods is to measure the oxygen content of the ice. We do this for a number of reasons, one being that the oxygen content gives you some idea of the elevation of the ice sheet surface (there is less oxygen the higher you go in the atmosphere; why it is so hard to climb Mt. Everest without supplemental oxygen tanks!). The other benefit is that melted and refrozen snow has fewer bubbles than normal glacier ice that began as snow and was compressed into ice with air bubbles. Where there are melt layers, one can expect lower oxygen content. We don’t see this at many places in the past, particularly in Antarctica. I believe at some sites in Greenland, such as the NEEM ice core collected in 2012, there is evidence of some melt during the last interglacial period ~120,000 years ago. Actually in 2012 the surface of the Greenland Ice Sheet experienced nearly total melt, producing a rare melt layer even at the summit of the ice sheet. These tend to be short, up to several day-long events, and aren’t large enough to destroy significant parts of the climate record preserved in the ice sheet.

      Another way we can double-check that we are not missing things in the past that may have melted away is to compare with other paleoclimate archives. Marine sediment records extend several million years into the past, and while they tend to be lower-resolution than ice core records they confirm the glacial-interglacial cycles seen in the ice core records, and we don’t see any anomalous discrepancies between these records. We can also look at speleothem–cave stalagmite records–to similarly check.

      Hope that answers your question!



      1. Elden Ferris

        Peter Neff.
        I’m way late to this site, however I find it enlightening as well as entertaining.

        1. the Oxygen content of ambient air is pretty much constant throughout the atmosphere. You need Oxy when climbing high mountains (or flying high) because of lower pressure. Our lungs need the pressure to absorb the Oxy.

        2. How precise are the dates gathered from ice-core.
        ( the reason for the question 1942 a flight of American aircraft landed on Greenland and are now over 300 feet into the ice. In some of the science involved in establishing dates in ice core, this would say that the planes have been there for 36,000 years)

  24. How recent can we read ice core data? I see that ice core data ends back in the 1800’s. I’m guessing that the ice has to compact for many years before it can be analyzed?

    1. Well, according to the first graph, we can read ice cores for about 10,000 years into the FUTURE, when it will be the year 2004.

      I was surprised, too.

    2. Bethan Davies

      Reply from Dr Peter Neff:

      Great question, Harold!

      We can read most ice core data right up to the present. We study both the air trapped in the ice (past atmospheric composition) and the ice itself (water stable isotopes for temperature estimates) and impurities in the ice (dust, salts, black carbon). For the ice itself, we can study right up to the present snow surface, and we usually do high resolution snow pit studies in the soft topmost few meters (hard to drill a snow core!).

      The primary limitations are for the gases trapped as bubbles in the ice, because—as Harold correctly guessed—this trapping process takes decades to centuries as snowfall slowly accumulates and compresses the air space between snowflakes enough to fully seal bubbles off from the atmosphere. At a place like Vostok, Antarctica or Dome C, Antarctica where the oldest records come from (400,000 and 800,000 years, respectively), there is very little snowfall—hence so many layers squeezed into the 3000m thick ice sheet—it takes a long time (several centuries) to fully trap bubbles of atmospheric air. In order to get younger gas records, scientists have gone to higher snowfall areas, especially Law Dome, Antarctica, where gases are trapped much more recently. This allows the ice core gas record to be connected with modern atmospheric measurements (more on that below). “Firn-air” measurements are even made in the top ~100 m of snow where the air spaces are still open to the atmosphere, to help understand how the gases are trapped in the ice and what age-range any trapped bubble represents (because the bubbles are open to the atmosphere until they close, any one bubble represents a sample taken over several/many years).

      Often, another fundamental constraint is how quickly you can go from drilling your ice core to publishing! It sometimes takes 2-3 years to get the cores, sample them, analyze the samples, interpret the results, write it up and get them published! In the case of deep ice cores, it takes up to 5 years or more to drill through the entire ice sheet before the full results can be worked on and presented.

      Here are two great resources for a little more on some of what I covered:

      A summary from David Etheridge, the Australian scientist who’s work at Law Dome connected ice core gas records with modern measurements:

      Unintentionally also from some great Aussie scientists, a brief overview of the whole ice core record up to pretty near the present:

  25. Thank you for this forum. What is the total uncertainty of the temperature measurements generated from analysis of ice core data? I realize it probably varies, but what is a ballpark number: plus or minus X degrees C?

    Optional bonus question: And how does this uncertainty compare to the temperature measurements of early thermometers, or modern thermometers? I ask because I’m trying to speak to a meteorologist in a politically conservative area about anthropogenic climate change, but he refuses to consider temperature data that precedes the advent of the thermometer because it was “not measured.” Thoughts? Thanks.

  26. Mario Buildreps

    There’s an obvious correlation between the Milankovitch cycles and D18O, although the feedback mechanism is troublesome, because the “lagging” varies significantly. The Milankovitch cycles don’t correlate very well with CO2 or CH4 and therefore not with any temperature reconstruction.

    1) How did you come to the conclusion that Milankovitch cycles are causing fluctuations in the CO2 and CH4 levels? Or is it still a suggestion?
    2) Why is eccentricity (Earth’s orbit around Sun) neglected as the sole correlating property with glaciation cycles (CO2, CH4)?

    Thanks, and good job.

    1. Eccentricity (Earth’s orbit around Sun) isn’t neglected as the sole correlating property with glaciation cycles. It’s the principle hypothesis for the glaciation cycles, I’m unaware of what any others are.
      Here’s the contributions assessed for a warming started 20,000 years ago and ended 8,000 years ago:
      The top-of-atmosphere (TOA) forcings/feedbacks of the most recent deglaciation were:
      0.5 +- 1 wm**-2 8% Orbital eccentricity, axial tilt & precession of the equinoxes changes forcing (what pulled the trigger that started it)
      3.5 +- 1 wm**-2 53% ice sheets & vegetation changes albedo-change feedback
      1.8 +- 0.3 wm**-2 27% CO2 change feedback
      0.4 +- 0.1 wm**-2 6% CH4 change feedback
      0.4 +- 0.1 wm**-2 6% N2O change feedback
      6.6 +- 1.5 wm**-2 total
      The theory is that this should lead to 6.6 / (5.24 * 0.61) * 220% (inc. feedbacks) = +4.5 degrees. I do not know why it actually ~ +5.5 degrees, some other 20% factor absent here.
      So above climate scientists have clearly stated that Milankovitch cycles are causing the glaciation cycles, and they’ve also assessed the principal feedbacks. What’s your issue with it ?

  27. Hi

    Great site!

    How do we know the age of air in the bubbles in the ice?
    It would seem that there would be a constant mixing of the air in the snow with “newer” air due to wind and air pressure changes. More importantly, as the snow compresses, “old” air would continually be forced up into newer snow layers. I could see the air in the bubbles being hundreds of years older than the ice it is found in, or mixed with newer air to the point of not representing anything.

  28. Great sit and great information on it.

    When low pressure is over a glacier, the air moves in the ice, up and out. When high-pressure is over the a glacier, air moves down and is pressed into the ice. Ice is not airtight until one or two hundred meter thick. (translates to thousands of years old in Antarctic ice cores) . So atmospheric air is clearly being gradually mixt in to the sample in question for at least some hundreds even thousands of years

    Based on what I’ve read here and beyond. I think it is likely that the ice core of CO2 measurements is at best hundred years average. Which means that a 10 year average could be much higher than the current 400ppm and the annual average could easily be up to thousands.

  29. Hello, It is great that you take the time to answer questions from non-scientists. Thank you for that. I am not a “Climate Change Denier”, but I am curious about certain patterns.

    I see comparisons between current climate data and climate data from various other time spans. These comparisons always show current conditions as “the most”, “the highest”, etc. For instance the greatest rise in temperature occurring in the past 150 years. Could that be due to natural warming after the “Little Ice Age” and returning to temperatures of the “Medieval Warm Period”?

    I look at the 800,000 year data, and it looks like at the start of all past intergalcial periods CO2 levels shot up rather rapidly as well, and nearly as high. Granted, not to 400 ppmv, but at least in one case to 300 ppmv. What caused those spikes in CO2?

    It has been proven that there is a correlation between CO2 and Temperature. Sometimes with Temperature leading, and recently with CO2 leading. But, I have not seen anything proving causation. Could there be another or even several other variables acting on both Temperature and CO2 levels causing them to rise and fall?

  30. Hello.

    This is maby allready answerd and sorry to revive an old thread but I have a question about icecores.

    Lets say that during a 50 year period 300.000 years ago, the CO2 peaked to 500ppm.
    Would you be able to detect this or would it thru the eons bleed of into the surrounding layers of ice due to the massive pressure from above?

    I do not deny we are heating up the earth but could you detect if so was made before?
    What is the “resolution” of the x-axis in these measurments.

    Hope someone sees this.

  31. If atmospheric CO2 rises after the temperature rises, then CO2 cannot possibly cause warming. Conversely, if CO2 is lower, biomass will suffer as CO2 is plant food. If biomass falls due to coldness, then the rise in CO2 surely must correlate with the lower absorption of that gas by the reduced biomass. The likely probability is that current rising CO2 is being driven by deforestation b man an a massive scale. The answer surely is reforestation on a massive scale. Only tiny Israel had more trees in 2000 than in 1900, due to a concerted effort by every government and a “Tree Planting” culture in the inhabitants that was created by this policy.

    1. Your information saying only Isreal has more trees in the last 100 years is not correct. Wealthy, mostly western countries in particular with increasing fossil fuel use have decreased and reversed deforestation. Deforestation continues where coal or gas power is not used for cheap effective fuel.
      “Large areas of the continent have seen a forest boom that means today more than two-fifths of Europe is tree-covered. Between 1990 and 2015, the area covered by forests and woodlands increased by 90,000 square kilometres – an area roughly the size of Portugal.”
      The planet has greened substantially over the last 35 years due to CO2 fertilization “The greening represents an increase in leaves on plants and trees equivalent in area to two times the continental United States.”
      This generation is living in a greener, cleaner and safer world.

  32. Ice age nett CO2 in the air reduces biomass, plants cannot live on ice. Plants absorb CO2 to make more plant. The rise in nett atmospheric CO2 AFTER the nett global temperature rises shows that global warming cannot possibly be due to CO2. Currently we are at the tail end of the last ice age, so naturally the globe is warming. Normally this would cause a rise in the earth’s biomass, but today that is falling due to massive deforestation by man for commercial usage. That rising CO2 is likely due to the reduced biomass, as well as the other non-man made factors like volcanoes on land and under the sea, the elliptical path of the earth’s orbit around the sun, the precession of the Eath’s spin and the passage of earth through the sun’s magnetosphere and solar wind.

    1. You are spouting a narrative rather than the data. Your thesis that “rising CO2 is likely due to the reduced biomass” is completely wrong. The CO2 increase is increasing biomass. Anyone who has understands carbon chemistry would know this is the expected outcome.

      1. Molecular biologist with over 20 years of research experience here. I doubt this ice core method is accurate as there’s no positive control. The only way you can prove this method has any merit is by comparing the CO2 in the bubbles to actual air samples from 400k years ago which is impossible. So it’s all just claims without proof. A lot of chemical reactions can happen over 400k years, affecting the CO2 concentrations.
        Not buying it.

  33. Can you inform me how you correct for the compression of the snow into solid ice sheets? We know that very little of the precipitation particularly in the Antarctic is anything other than crystalline ice so this is very loosely packed. This therefore would require enough of a build-up over time to collapse it down into solid ice. This would also require the proper weather conditions. While I expect that it would average out over time, wouldn’t this change the actual position of peaks and valleys in the data? I would think that you could determine large scale timing by matching dust particles with geological evens such as extreme volcanic or meteoric events but small scales may be important.

  34. it depends a lot on how much snow falls! Where you have tens of centimeters of snowfall each year you can generally see the annual layers if you know what to look for. We sometimes dig pits into the recent snow, and with back-lighting, you can see layers–google “Antarctic snow pit” for some great examples. Based on how the winter versus summer snow “packs” the layers are distinguishable (windier in the winter means smaller broken bits of flakes pack more densely, I believe). Also, in Greenland ice cores the layers tend to be very easy to see because there is much more dust in the atmosphere in the Northern Hemisphere summer (when so much land area becomes snow-free). These dusty layers are darker. Once you get a few hundred meters deep, the ice becomes very glassy and more homogeneous to look at, in my experience, but under the right light conditions, experts can still see the layers. Thankfully, we can much more easily “see” annual layers in the chemistry of the ice and can count back ~40,000 years.

    1. Chris Schoneveld

      I thought you could count back 450000 years (in the Vostok core). I don’t get it.

  35. If atmospheric CO2 rises after the temperature rises, then CO2 cannot possibly cause warming. Conversely, if CO2 is lower, biomass will suffer as CO2 is plant food. If biomass falls due to coldness, then the rise in CO2 surely must correlate with the lower absorption of that gas by the reduced biomass. The likely probability is that current rising CO2 is being driven by deforestation b man and a massive scale. The answer surely is reforestation on a massive scale. Only tiny Israel had more trees in 2000 than in 1900, due to a concerted effort by every government and a “Tree Planting” culture in the inhabitants that was created by this policy.

    Reply ↓

  36. Your “If atmospheric CO2 rises after the temperature rises, then CO2 cannot possibly cause warming” is an obvious logical fallacy and non sequitor as any person who is aware of the basics of “feedbacks” knows. It’s the exact equivalent of confidently and incorrectly stating that if Iggy Pop screaming into a microphone causes sound to come from his 99 loudspeakers after he does that (i.e. if his dulcet tones don’t magically exit the speakers prior to his instrument opening), then loudspeakers placed next to the microphone cannot possibly cause any more sound to go into the microphone and come out of the loudspeakers again. Here’s the contributions assessed for a warming started 20,000 years ago and ended 8,000 years ago:
    The top-of-atmosphere (TOA) forcings/feedbacks of the most recent deglaciation were:
    0.5 +- 1 wm**-2 8% Orbital eccentricity, axial tilt & precession of the equinoxes changes forcing (what pulled the trigger that started it)
    3.5 +- 1 wm**-2 53% ice sheets & vegetation changes albedo-change feedback
    1.8 +- 0.3 wm**-2 27% CO2 change feedback
    0.4 +- 0.1 wm**-2 6% CH4 change feedback
    0.4 +- 0.1 wm**-2 6% N2O change feedback
    6.6 +- 1.5 wm**-2 total
    The theory is that this should lead to 6.6 / (5.24 * 0.61) * 220% (inc. feedbacks) = +4.5 degrees. I do not know why it actually ~ +5.5 degrees, some other 20% factor absent here.

    Your “CO2 is plant food. If biomass falls due to coldness” is a massive red herring because I ate well 60 years ago with far lower CO2 and I recall we had trees & flowers. So, since natural CO2 reduction cannot possibly reduce CO2 back to where it was when I was a kid (natural CO2 reduction recently reduced it a massive 720 ppmv over just 47,000,000 years) then unless you intend to be much more of a big eater and a fatty than me as kid (almost impossible) it is a non issue for a few thousand years at least so I consider it low priority. Are your priorities to start by worrying about the possible problems starting in a few thousand years ?

    1. Your analogy about eating well 60 years ago misses 2 important points. Firstly the expression greenhouse gas was first used to describe CO2 around the late 1960’s by the environmentalists of the day to establish a cause. Secondly it was chosen because commercial greenhouses and nurseries pumped CO2 into the greenhouses to encourage growth. This still happens today and is why you had, and we still have, plentiful food flowers and trees.
      Incidentally the CO2 neither heats nor insulates the greenhouses.

  37. I am a molecular biologist and have to say none of this ice core ‘science’ is convincing. I don’t see a single control I would call reliable. First of all, you may have chemical reactions within these enclosed bubbles that change CO2 concentrations, for example formation of H2CO3 and HCO3- , resulting in lower CO2 levels. None of this would be detected in the air you analyze. Many other potential chemical reactions could take place that change the CO2 concentration in the enclosed air. Furthermore, microorganisms such as algae or anaerobic bacteria enclosed in the ice could metabolize CO2, resulting in lower CO2 levels as well. Just because they are frozen doesn’t mean their metabolism is at an absolute standstill. As a matter of fact, they still metabolize. Most frozen microorganisms most likely won’t survive more than a few thousand years. Under lab conditions, cryopreserved cells at much lower temperatures already go bad after after 10 years, now compare this to 400k years. The cells would lyse and you would be unable to detect them. Even if some were still some alive, their numbers would be much lower than initially and the impact of these microorganisms on CO2 levels in the bubbles would be underestimated.
    Taken together, this would result in CO2 concentrations that are closer to the surface.

    Furthermore, your posted Vostok ice core figure is laughable. The last data point on your graph is still within normal range relative to the previous measurements on the left. Then the graph suddenly reaches new heights, going from ca. 280 ppm to 377 ppm in 2004. I doubt ice core data less than 15 years old is comparable to ice core data 400k years old. Also, please don’t tell me you measured 2004 CO2 concentrations in the air and then directly compared this to your ice core data. These are two entirely different methods and, no matter what climate or ice core scientists claim, their results will not, never ever, align properly or can be directly compared since you derived them using two entirely different methods. This is like a molecular biologist measuring a protein concentration using ELISA and then appends these results to protein concentration measured by the Bradford assay. No scientist would ever do that, yet it seems climate ‘scientists’ seem to get away with it. It’s simply baffling.

    1. I think a good term for it is “negligent rationalization”. They’ve already decided what they are going to believe, and they’ll make sure that their data supports what they’ve already decided to believe.

  38. The math seems very off to me. I understand climate change, we go through warming and cooling periods.
    the levels don’t make sense, the post 1950’s CO2 levels seem exaggerated. If true only ~30% of the CO2 is absorbed in the ocean, then only ~30% of the CO2 would be trapped in the ice core samples.
    That would leave ~70% unaccounted for; Yet it seems that the measurements today are taken form the atmosphere. that ~40% difference in the ppm.
    Can you explain how you get the avg. temperature from the core samples, this is my biggest doubt, it would be equivalent to using an ice cube from my freezer to guess the temp. in New Delhi.

  39. Thank you Bethan Davies and all of the other researchers and scientists for this very very interesting article.
    Before reading this and several links provided I was a bit sceptical on whether C02 and other GHG’s were causing global warming, I was completely ignorant of the Ice core data actually proving this point.
    My question then to you is, with the GHG’s already released in our atmosphere twice as high as prior to the 1800’s, and C02 which is trapped in the Ocean and atmosphere, (I read it in one of your articles that it will now stay trapped for between 500 – 1000 years) what affect will this alone have on climate change if we were to stop releasing GHG’s right now?
    I have so many questions, this article has opened up my mind to Climate change. Thank you for posting this.

    1. Unfortunately you misinterpreted the data and may not of read all the information. The Ice core samples show that temperature increases, then CO2 increases about 800 years later. CO2 does not lead temperature up, it lags it. So the conclusion you should come away with is CO2 does not cause global warming.
      Petit et all 1999 — analysed 420,000 years of Vostok, and found that as the world cools into an ice age, the delay before carbon falls is several thousand years.
      Fischer et al 1999 — described a lag of 600 plus or minus 400 years as the world warms up from an ice age.
      Monnin et al 2001 – looked at Dome Concordia (also in  Antarctica) – and found a delay on the recent rise out of the last major ice age to be 800 ± 600
      Mudelsee (2001) – Over the full 420,000 year Vostok history Co2 variations lag temperature by 1,300 years ± 1000.
      Caillon et al 2003 analysed the Vostok data and found a lag (where CO2 rises after temperature) of 800 ± 200 years.
      NOAA data on Antarctic Ice Cores:

  40. Hello and thank you for sharing your expertise.

    I have a very smart friend with superb memory who doesn’t believe in climate change which makes it really hard to debate with him (he’s a much better debater than me). He contends that many of the ice core samples which are used to demonstrate the low levels of CO2 over vast expanses of time have been stored incorrectly, allowing the CO2 stored in the cores to ‘leak’ into the environment and thus showing dramatically lower CO2 levels over past history than is fact. He says he has visited a few ice core facilities and has seen this; and contends that the cores need to be stored extremely cold to have any accuracy. While I understand what he is saying, and even acceding that perhaps some cores were stored incorrectly, I cannot believe at this point in climate change studies that this would be a significant factor. There must be hundreds of core studies by different trusted research facilities and institutions which properly store the samples so that we can have confidence in the data being shared (such as your graph showing the rise of CO2). Can you help me address his points or point me in the right direction?

  41. I was wondering what the expected Margin of error is when measuring previous years temperatures through ice cores?

    Is it expected to be + or – 1 degree celsius?
    0.1 degree? Or is it more like + or – 5 degrees?

  42. Nguyen Le Anh

    Dear Bethan Davies,

    Thanks for the explanation.

    Davies, you have said that air bubbles trapped in ice cores have been communicating with air for years, even 1000 years. If so, how are studies on the earth’s climate within a few thousand years, or a few hundred years?

    I do not know if the greenhouse gas model is correct enough for a long time of many hundreds of thousands of years to determine the temperature through the presence of some gases.

    Could you please explain to me more clearly how does the information obtained from the ice core give a claim about the sea level?

    Best regards

  43. Craig Marshall

    Here are some thoughts on lag time: It is very challenging to put CO2 records from ice cores on the same timescale as temperature records from those same ice cores, due to the time delay in trapping the atmosphere as the snow is compressed into ice. It is my understanding that the ice at any time will always be older than the gas bubbles it encloses, and the age difference is inherently uncertain.
    Air is trapped in a layer only after the snow above it has built up to a thickness of 70 meters or more, and the time this takes can vary greatly as the climate changes.

    I understand that Skeptics argue that since CO2 lags temperature, so CO2 rise is a response to temperature change and not the cause of it. I also understand that CO2 didn’t initiate warming from past ice ages but it did amplify the warming.  In fact, about 90% of the global warming followed the CO2 increase.

    Is my understanding on the right track?

    Here is something I don’t understand: In the Antarctic core record, how is it possible that the turning points in the temperature record occur at maximum CO2, i.e. cooling phases begin when CO2 is at a maximum. It does seem odd that when CO2 levels are at the highest, temperatures start to fall and Earth eventually has advancing glaciers.
    I understand that at the beginning of an interglacial, the Milankovitch Cycles (along with CO2, albedo, and other feedback loops) nudge the Earth out of an Ice Age into an interglacial? Does the same thing happen in reverse at the end of an interglacial?

    1. CO2 is not a primary driver of temperature, it is a bit player. As CO2 increases it has less warming effect. Once the concentration of CO2 in the atmosphere gets past about 250ppm has barely any effect on temperature. Other climate drivers overwhelm the effect of CO2s role, so CO2 can be at 1000ppm and the planet will still cool due to changes in obliquity and other factors.

  44. Nick savino

    so how do you know that layers form by year and not by accumulation rates? if you are going to count them like tree rings that is not a good method because there is a lot of things that can interfere with that, such as random snow melting, which would turn into water, and then that water freezes which forms another layer

    1. Laura Boyall

      Hi Nick,
      Good question. We know that layers are annual as there is typically a change between summer and winter ice, which we can visibly see and count. In addition to this, we can use marker horizons such as tephra layers. Most of these have a known date of deposition, therefore we can calibrate our counting.
      I hope this helps,

  45. I’d like to know how do you stop the sun from reflecting down the hole that is left once an ice core sample is taken?

    1. Laura Boyall

      Hi Leah, Thanks for that question, it is really interesting.
      The ice flow is constant so it acts to close up the hole quickly. This means that the sun won’t have much of a chance to reflect down the hole.

      I hope this helps 🙂

  46. I cannot read very good, and have problems understanding every thing said,
    Can someone please help me to understand the following.
    Is there any layers of thick ice with a cyclical periods and if so how thick is the thickest ice and how often do they occur?
    Could you please email me an answer as this is the only way I can learn and receive understanding with the hep of others,
    Thank you

  47. Chris Schoneveld

    Do I take it that the vertical (time) resolution is one year? Can one distinguish 450000 layers in the Vostok core, which spans some 450000 years?

  48. Hi Bethan
    Thanks for the site information. This is important work. I don’t doubt the accuracy of the mass spectrometry measurements of the gas removed from the cores.
    The most recent measurments line up well with actual CO2 measurements, but going back in time wouldn’t gas loss/exchange/degredation cause an a lower measured CO2 content than what existed in the atmosphere at the time?
    Although the cores show helpful relative temperature and CO2 measurements. Would you agree that the ice core samples provide a lack of fidelity and smoothing of gas concentration measurements compared to plant stomata proxies?
    See the research referred to here

  49. By the gradual melting of the glacial ice, due to global warming, some countries are falling into loss, while for others it is very beneficial, like Russia. Most of the seas around Russia are frozen, and it is almost impossible for Russia to keep on its trade all through the year. But now, there is hope for Russia, of opening these seas for international trade. If we review all the glacial ages, the temperature did not remain the same all through the glacial age. At a stage of a peak, the temperature had fallen to an alarming point, but as the large part of the glacial age passed, the temperature became optimum. I think the recent era is the time of optimum temperature, which can be beneficial for all countries, even for the northernmost and southernmost. Res:

  50. I have a several questions for an expert, or experts, on CO2 measurements taken from surface glacier ice in Antarctica. I am a former petroleum goelogist working on a book regarding certain events during the onset of the Younger Dryas.
    Specifically, if samples (~500 g) were taken from surface cores of ice dating to ~13 ka to ~12 ka, and the lab measured CO2 levels were in the 250 ppm range, what is the likely error range for those CO2 levels (in SD or SEM)?
    What specific things are happening in the ice, post-deposition, that may have impacted the measured CO2 level?
    What is the typical thickness of ice used in the CO2 measurement? Presumably this sample “thickness” represents many years of time. Can we say how much? Therefore what is the typical temporal resolution of a surface glacial ice core CO2 measurment?
    I’m looking for a “champion(s)” on this topic. To date I have interviewed 29 PhD’s for my upcoming book. I would certainly include your thoughts in my book especially if you have some academic or research credentials. Thank you. Eric
    Please reply to if this is allowed.

  51. Looking for a champion to help me with questions on my next book.
    – If we took a surface glacial ice sample (13-12 ka old) for Antarctica, analyzed for CO2, and obtained a measurement of 250 ppm, what is the margin of error for this CO2 measurement (SD or SEM)?
    – What post-depositional events can change the actual atmospheric CO2 level at time of deposition?
    – Any thoughts on the temporal resolution of (age of deposition range) for glacial surface ice measurments taken from Antarctica?
    I have interviewed 20 PhD’s to date for my book so if you have suitable academic or research credentials regarding Antarctic ice CO2 measurments, I would like to speak with you regarding my research. Kind regards, Eric.
    Email me at

  52. The charts, no matter what they are measuring in the ice core data, seem to spike every 12,000 to 13,000 years. Could someone please measure the average number of years between spikes and their statistical variation from the mean? These spikes appear to be in sync with the half cycle of the precession of the equinox. [By the way, precession is not due to an earth wobble but to a helical movement around a central axis (i.e.; galactic orbit) with our solar system juxtaposed to the Sirius system.]

  53. This article is very interesting and I love it! yall should do one on woman’s history about women who have not been credited for things they did . for woman’s history month. ????????

  54. Seems to me that to get readings for today’s levels that are comparable to the historical readings, we need to wait a long time until the current surface is 50-100m deep, as it was mentioned that this is how long it takes for the air bubbles to become trapped in the ice. Whether that is really a reflection of the CO2 levels from when that stratum was at ground level or of the levels just before the air got trapped, is maybe not as relevant as the necessity of comparing like-for-like.

  55. John Krainski

    Thank you for your many years of intelligent and thoughtful questions on this site. In my admittedly amateur efforts, I’ve found the number one issue that people get stuck on is the 800 year lag of CO2 to Temperature in the ice core data. My understanding of the current theory is that: 1) Milankovitch cycle results in more solar irradiation of Earth’s surface, 2) resulting in initial global temperature increases, 3) these temp increases, although small, have an impact on ocean temp which drives release of CO2 due to reduced solubility in seawater, 4) there is a significant lag in this CO2 release because oceans take a long time to heat up, 5) this appears to be the primary source of additional CO2 in the atmosphere, 6) additional atmospheric CO2 greenhouse effect results in even more global temperature increase than would have occurred due to solar irradiation alone, 7) there are some built in limiters that eventually reverse the process.

    First, can you tell me if I’m understanding this correctly? Second, can you point me to a study or model that ties the basic physics of a) ocean heating and b) CO2 as a greenhouse gas to the relevant times. In other words, is there a model that illustrates the time to heat the ocean, the change in CO2 solubility, and the resulting CO2 absorption by atmosphere and resulting temp increase?

    I’ve read through all the comments so far (and searched far and wide) with no straightforward results yet. Thank you in advance for your time.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

This site uses cookies. Find out more about this site’s cookies.