Climate Science for Everyone: How scientists measure the carbon dioxide in 800,000 year old air

Update: To read other articles in this series, click here.

Climate scientists who study the history of the Earth’s climate (also known as paleoclimatologists) know that modern carbon dioxide levels are at their highest level in the last 800,000 years. They tell us this because they’ve been able to measure the carbon dioxide in air that is actually 800,000 years old. So how do they do that?

Scientists know how much carbon dioxide was in the air hundreds of thousands of years ago because they actually have small samples of ancient air stored in glacial ice. To get a feel for how this works, consider the following examples.

Go to your freezer and pull out a few ice cubes. Look closely at the cubes and you’ll see bubbles frozen in the ice cube. My ice maker produced easily visible large bubbles (~ 1/8th inch across) in about every third cube. Those bubbles are filled with air, so it’s pretty obvious that ice can capture and store air. If you put that ice cube in a refrigerated time capsule for 800,000 years, some future scientist could pull it out and they’d have a sample of the air from your freezer that was 800,000 years old. The same thing happens naturally in lakes, as these frozen bubbles of methane show. But what about snow?

First, we know that fallen snow contains a lot of air – that’s why it’s so light when you shovel it. Even wet and heavy snow has a lot of air, as you can feel by how much you can compact it as you’re making snowballs – compacting the snow into a ball squeezes the air out of it. Even at the bottom of a deep snowdrift there’s still lots of air mixed in with the snowflakes. As snow keeps falling, eventually there is enough to keep the air several feet down from mixing with the atmosphere. When that happens, the air is trapped and essentially becomes a time capsule, storing a sample of air in snow much like that air bubble in the ice cube did. The trapped air is released back into the atmosphere when the snow melts.

Second, anyone who lives where it snows several inches in winter already knows that snow eventually turns to ice just by compressing it – it’s how snowy roads turn icy just by cars driving on them, even when the weather is so cold that the snow doesn’t melt. The same thing happens in a glacier or ice cap, but instead of cars compressing the snow into ice, the weight of years or decades of accumulating snow is what turns old snow into ice. And when the snow turns into ice, the trapped air forms small bubbles in the ice.

Scientists can extract the air from those bubbles and then use instruments (namely mass spectrometers) to directly measure how much of what gases are present in the air. The depth of the ice from the surface tells scientists how old the ice is. And from this information, scientists can plot a graph of the concentration of a gas (such as CO2, or methane) over time.

When scientists do this, they create images like the one above from the 2007 IPCC “state of the science” Working Group 1 report. Note those stars in the upper right corner – they’re the “present day” measured concentrations of nitrous oxide (green), methane (blue), and carbon dioxide (red). Note also that they’re on the same scale as their respective lines – so the blue methane star (1750 parts per billion, or ppb) is on the same scale as the blue methane line, which never gets above about 800 ppb concentration in the ice core record. Similarly, the red CO2 star (370 parts per million, or ppm) is on the same scale as the red CO2 line, which never gets above about 300ppm in the ice core record.

So why does this matter? Well, because it means that scientists can use air trapped in snow from hundreds of thousands of years ago to tell us how much of each greenhouse gas there was in the air back then. And that gives us a baseline to compare modern concentrations of greenhouse gases to old concentrations. And that tells us that CO2 and methane haven’t been this high in hundreds of thousands of years, even as ice ages have come and gone.

In the next installment, we’ll learn about how scientists can use these same ice cores to estimate how the global average temperature has varied over the last several hundred thousand years.

Image Credits:
Katey Walter

26 replies »

  1. What is the differential diffusion of CO2 through ice under all the pressure, shear, temperature fluctuations over hundreds of thousands of years, vs oxygen and nitrogen?

    What is the CO2 component in the snow itself as it fell? What are the equilibria concentrations of all three of these species in both the bubbles and the ice under all conditions and times?

    Are you sure?

  2. One simplification:
    “..As snow keeps falling, eventually there is enough to keep
    the air several feet down from mixing with the atmosphere.
    When that happens, the air is trapped and essentially becomes
    a time capsule, ..”

    seems to totally ignore the issues that:
    1) the entrapment is when the snow changes to ice – about 70-150 years
    2) the core is initially very brittle and they let it “relax” for a year
    3) stress and cracking smears large vertical areas.
    4) we have no clue about the diffusion of different gases within the snow lattice

    But apart from huge error bars and the data showing that warming releases CO2 (trends after), it is all excellent eye candy.
    And besides,
    once China gets their moon base and can throw
    “environmentally friendly” rocks at us,
    we’ll really have to repay Grandpa Baby Boomer’s
    “borrowed our way to prosperity”!

  3. LOL in Oregon, how does any of that affect the physical mechanism I described, or change the fact that mass spectrometers are used to measure the carbon dioxide?

  4. Brian,
    Nope, doesn’t change the mechanism or the outstanding effort put into the work to pull the cores and experimentally measure what the machines measure.

    The question is how can it be interpreted.

    We don’t have a theory to explain the delta in the numbers, we just have numbers with huge error bars and we don’t even really know the magnitude of the error bars since we have insufficient data samples for each micro period and we have zero calibration samples from 700kya and must “assume”, from extremely short experience, what the numbers might mean.

    Drawing conclusions to match the funding agenda du jour is the same as standing
    atop Mt. Stupid and waving the flag of “this must be it”.

    On the other hand, we still have to get passed the Me, Me, Me, evangelical Baby Boomer generation before “The Shallows” of the younger generation gets to display the superficiality of their excellent multitasking/no thinking methodology.

  5. A bit of a coincidence that they all should go up and down together, don’t you think LOL of Oregon? And if there were any significant diffusion, why are those sharp terminations still there (and equally sharp from the beginning to the end)? And why the match with the astronomy (grey bands)?

    You don’t have zip mate. And don’t try to bluster your way out — science is more than using multisyllabic words.

  6. This is a very helpful post about CO2. I had to answer a question like this for a school project. This has helped me a lot.

  7. Reblogged this on Weblog at the edge of the world and commented:
    The carbon isotopes do not lie; the facts in evidence by the isotope ratios reveal that the current rise in carbon dioxide is due to combustion (heavy carbon is produced by high-energy combustion) and not a “natural” process.

  8. I have a question about these ice cubes. The article mentions 800,000 year old ice. Ok. Do we have ice samples dated back to 700,000 years ago? Or 2 million years ago? Or 50 million years ago? I ask this because if we don’t have a huge sampling of ice from a vast number of time periods, then how can we know what the CO2 levels were in between the ice samples we have? This is simplistic, but if we have an 800,000 year old sample and a 700,000 year old sample, what can we know about the 100,000 years in between? Do we use computer models to estimate? Is it possible CO2 could have gone way up for 200 years in between those two sample? If it did, is there any way to know it did? Or any way to know it did not? I’m trying to understand not just how we know CO2 levels at the time of dated ice samples, but how we can be sure about CO2 levels in the time periods in which we do not have ice samples.

    • Yes, we have continuous CO2 records for the last 800,000 years. Different ice records from different parts of the world go back different times – some records go back 300,000, others 100,000, others 600,000.

  9. Thank you, but let me seek clarification. When you say continuous, do you truly mean continuous? In that, for 800,000 years, there are no missing data? No gaps at all? That’s very important, for any information gaps at all leaves room for questioning the claims that CO2 has never been over 300ppm. If I make that assertion, I need to be absolutely sure there’s no room for error. Even data missing for just a 100 year period in that 800,000 year span leaves a small opening for critics. If that opening is there, I want to know it’s there. Thanks again.

    • So basically the ice record is continuous, but due to practical limitations of slicing small amounts of ice and extracting the air bubbles (and how much trapped air is needed to get an accurate measurement), scientists can only get measurements every few decades or centuries. And because of ice compression, the further back we go, the coarser the measurements get.

      • Thank you for taking the time to clarify. Those are important details to know and I wish more people on both sides of the argument would be up front about what we actually know and don’t know. We might spend more time making progress and less time arguing.

        May I ask one more question of you? What explanation is there for the latest NASA data (I’m speaking about pictures posted on their website) that shows Antarctic ice actually increasing in the last few years? I realize the complexity of underwater currents and temperatures makes this difficult, but when someone says ‘yes, ice is decreasing in the north but it’s increasing in the south’ how do you explain that? It’s frustrating because most climate change sites and blogs avoid the NASA data instead of dealing with it, which makes them look like they’re cherry picking their data.

        • Jim, my answer to your question depends on whether you mean Antarctic sea ice or Antarctic land ice.

          There is a key difference between the Arctic and the Antarctic that results in very different responses at the two poles to global warming: the Arctic is an ocean surrounded by land, while the Antarctic is land surrounded by an ocean. Sea ice around Antarctica is largely controlled by how fast glaciers are emptying into the ocean, while sea ice in the Arctic is largely controlled by how cold the air gets and for how long.

          Scientists have relatively recently discovered that currents in the Southern Ocean are warming up a little and are eating away at the bottom of ice shelves and glaciers whose bottoms are below sea level. This results in reducing the inertia of ice shelves and reducing the friction between glaciers and the sea bed, increasing the rate at which ice from the interior of Antarctica can flow into the Southern Ocean. Faster ice flow equals more (but usually thinner) sea ice.

          The Arctic works differently. Since sea ice is mostly a function of cold winter air temperatures, anything that increases air temperatures works to reduce sea ice. Since the temperature of the land around the Arctic Ocean increases faster than the temperature of the ocean itself, warmer air for longer in the autumn leads to a shorter ice-forming period, lower maximum ice extents, and thinner ice.

          There’s also a major difference between the color of the Arctic ocean vs. the color of Antarctica. There are precious few places in Antarctica that aren’t covered with ice year round, and so it’s always colder due to the high reflectivity of snow and ice. This means that the amount of solar energy absorbed by Antarctica is quite low, depending on the age of the ice and snow (30-85% reflected). In contrast, open ocean water absorbs nearly all the solar energy that hits it (<10% reflected), which slowly warms up the water and makes it harder to cool down to a freezable temperature the following winter.

          Finally, one NASA study in 2015 found that total ice over Antarctica was increasing, but multiple studies using two unique gravity measuring satellites disagree (GRACE and GOCE). Most of the scientists I know think that the gravity measurement-based approaches are less sensitive to error and thus more likely to be correct, but I haven't seen much on the subject in the last couple of years (although I haven't been looking really closely). At this point there's disagreement about which approach is "better" and where the disconnect is happening (since one or both approaches have to be wrong in some way). I'm actually happy that there's not been much on the subject from serious scientists, because that means they're working to figure out what's really going on, and that's how science discovers the truth.

          The Skeptical Science page on the Zwally et al 2015 Antarctica study is a bit old, but decent nonetheless: https://www.skepticalscience.com/argument.php?a=21&p=4

  10. Could an answer be put up to tue question made on Jan. 28, 2012?

    Also, I am wondering separately what kinds of percent error exists in the carbin dioxide calculations due to the unknowns in that question and how much percent error is there in the temperature calculations for the same time periods hundreds of thousands of years ago or millions of years ago.

  11. So based on comments it sounds like the scientific community believes we have climate change occurring that is due to greenhouse gases, however the scientific community is relying on inconclusive evidence ie. no proof that CO2 levels may have had similar levels at sometime in the past as it is today. We need to study old ice with samples taken for each 100 years going back to 1,000,000 years, that would be conclusive evidence. Everything spouted about today is false until this is done.

    • Essentially what you’re saying is that, because we don’t know everything, we don’t know anything. There is an overwhelming amount of other physical evidence that you’re ignoring, and I can’t help but think that even if scientists gave you exactly what you are asking for, you’d shift to say that we need resolution of 50 years, or 25 years.

      You’re being unscientific.

  12. Brian someone asked about the potential error in the analytical process and the sampling. All analytical methods have an error issue but with the samples being what they are the interpretation is more problematic. Scientists? are looking at numbers and making an interpretation that these numbers are “Perfect” which is not true. There are many instances where interpretations have been made and the actual outcome does not tally with the initial results. Sea lions dying?? populations reducing but in fact they have risen significantly. It would seem the money to be made out of Climate change is more important that the actual story itself. When you describe the ice core samples it would seem that they are not somethiing that we can say in “1786” the level was 0.04ppm CO2 but we can “indicate” that in 1700 -1800 the level was approximately 0.04ppm +/- “what error?” in this period of the planet people talk about the death of our species in 12 years time and the only reason is to actually destrioy our species. Not sure why?? option a) plant millions or billions of trees option b) have a war and reduce populations by 50%?? INSTEAD OF DOING AWAY WITH ALL INDUSTRY , REMOVING CATTLE AND METHANE PRODUCING ANIMALS FROM THE PLANET AND LASTLY TELLING SCHOOL CHILDREN A WHOLE LOT OF CRAP TO STIMULATE MORE RESEARCH DOLLARS??

    • In my experience, scientists don’t ever assume the numbers are perfect as you suggest. For example, the following papers make explicit estimations of the error in their CO2 ppm measurements from their ice core data:

      Evidence for substantial accumulation rate variability in Antarctica during the Holocene, through synchronization of CO2 in the Taylor Dome, Dome C and DML ice cores has an estimated 1 sigma error using an analytical method of 1 ppm.

      Termination V in the Vostok (Antarctica) ice core estimates 1.5 ppm for EPICA Dome C and 3 ppm for Vostok.

      Keep in mind that scientists are able to project certain effects with good probability, but predictions are not possible. And the difference between a projection and a prediction. Furthermore, no-one who has studied this field even somewhat seriously expects that the Earth’s mean temperature will increase steadily without any dips or pauses. There are too many other factors that affect the climate on short scales (El Nino, volcanic eruptions like Pinatubo, and even the 11 year solar cycle) to predict what will happen with one region or one species (sea lions, for example), but that doesn’t mean scientists can’t project what will happen over the long run.

      Also, I’ve researched where the money is, and it’s definitely not in opposing climate change. Read this article if you’re interested in learning more.

      • The money is in the area of proposing climate change exists and that in the next few years the planet will beconme a waste land (or Sea) and Mankind will not exist.Your error statements are reasonably low is that for the process of sampling and analysis or in the analysis only ? what are the controls used to estimate the results and is there any controls that would represent the samples from an ice core??

        • Read the papers and their references if you want to know how they did it. That’s why I provided the links.

          There is no “if” with respect to the fact that climate is changing, it’s due to industrial emissions of carbon dioxide, and that those changes will be disruptive to the biosphere and human civilization. The question is how bad will it get and how fast. Human civilization won’t collapse in 12 years, but after that, really bad effects are all but locked in.

        • Brian Thank you for your answer but it did not address the controls used within the analytical process that should also include the sampling. a brief summation would be nice. Do you not think that these scientists should be providing answers to the problem not just saying there is a problem. In Australia we have banned Nuclear power yet as an alternative to coal id would provide base load power and NO emmisions?? As far as 12 years that is the period that the young children have been given and as such have created a massive issue with school children instead of them provideing answers for school children. Our prime minister is currently visiting the south sea nations where it has been stated their islands are sinking yet this is a lie since 8 out of ten are actually growing. They are calling for Australia not to open any new coal mines,yet our coal is significantly cleaner than most of the rest of the world.Can’t these issues dealt with by intelligent thoughts not Panicing everyone.??