“Historically, the claim of consensus has been the first refuge of scoundrels; it is a way to avoid debate by claiming that the matter is already settled.”
- Michael Crichton

You don’t tend to hear too much about ice-core data these days, which I find interesting.  When I was working as an atmospheric chemist in the mid to late 1990s, ice-cores were *the* main focus of those who were trying to understand past atmospheric compositions.

People trying to figure out the Earth’s radiation budget (flow in – flow out = net heat gain or loss) have to care about CO2.  Of course they’ve only been worrying about these things for the relatively recent past, and only able to measure them both simply and accurately since the 80s.  Some of the more reactive species of importance in the atmosphere still can’t be measured accurately even with fiendishly complex modern equipment.  So the instrumental record of directly measured carbon dioxide in the atmosphere is limited to the last 60 years.  No doubt scientists could have measured CO2 before that, but no one really saw the point.

So we start to realise that atmospheric CO2 could be a big driver in Earth’s climate, but we also know climate has changed dramatically over past decades, centuries and millennia, but obviously we don’t have CO2 measurements from back then.  But then someone with some knowledge of glaciology realises that its pretty easy to get to very old ice in various parts of Earth, and that ice has air bubbles included in it.  If that air was the same age as the surrounding ice, and you could extract and measure the CO2 concentration in that air bubble, that would allow you to determine the CO2 concentration WAAAY back when the air was trapped.  Everyone knows ice flows, so if you know how fast it’s moving, you can basically figure out how old the ice in a particular place is. Bingo, now we can measure CO2 levels back in time to the age of the oldest ice on the planet.

Sound too simple?

It is. People started looking at ice core data for things that are known to only be man-made (like CFCs) and comparing their concentrations from a known age of ice core to actual measurements made at that time.  They didn’t agree.  Once you start to think about it there are a LOT of assumptions in ice core data.

That ice started out as snow.  Anyone who’s played in snow knows it’s pretty porous.  There’s a lot of air in snow.  All that trapped air gradually gets squeezed out as the snow thickness accumulates. Eventually the snow becomes ice, and the air is finally trapped.   But up to that point, how do you know what the rate of exchange of air in the snow is with air in the atmosphere? How long it takes to seal off the firn and turn into ice depends on how much snowfall there is.  Which might be known now at the site, but how would you know that rate at that site from several centuries ago? Or from several tens of centuries?

The timescales for sealing off firn are long: several years where snowfall is light.  That’s a long time in a physico-chemical sense where most things are measured in minutes or hours at most.  Gravity becomes important over that timescale and heavy molecules tend to concentrate at the bottom of the firn as light ones tend to move towards the top.  How do you calculate the effect that this gravitational settlement has on the composition of the trapped air?   Strong winds can still flow through the firn, further mixing the air.

The ice gets buried in more and more ice, but it’s also flowing, and if ice flows fast, eventually it’s going to get to somewhere warm enough to melt, and then your air is released again.  So for a long timeseries of air from an ice core, you need to drill a core where the ice flow rate is slow.  This happens at ice ‘domes’ tops of hills from where ice gradually flows away.  But as the ice build up increases, so does the pressure, so age is not directly proportional to depth.  1 cm of core at the bottom, represents a lot more time and ice accumulation than 1cm at the top.  And wherever you drill, you have a long time condensed into a tiny thickness of ice.  The exact age-to-depth conversion is tricky. You’ve got to consider the terrain, the precipitation rates over the lifetime of that core, the mass flow rates (which depend on what’s happening at the top, and at the bottom of the ice sheet).

Then there’s the clincher.  You have to assume that nothing happened to the air in that bubble during all that time it was sitting around in the ice.  Nothing is on your side here.  CO2 might be regarded as inert by atmospheric chemists, but the biologists know it’s a fundamental part of most biological processes.  Anything that uses or makes CO2 will alter its concentration in the ice.  We used to think ice fields were pretty sterile places.  Now we know life is so tenacious it seems to be everywhere we’ve thought to look for it.  Granted, in ice life moves at a slower pace.  But in ice cores, time is one thing you’ve got plenty of.  We now know that the firn is a veritable photochemical chamber.  All kinds of photochemical reactions are going on.  And we know all air has stacks of microbial life in it. Microbial life that will be intimately associated with CO2.

Then there is the ‘container’.  Most of the atmospheric chemistry that has been looked at is ‘gas-phase’.  That’s mainly because its relatively easy to study gas phase reactions.  But we know that cold ice is a fantastic place to do a whole load of chemistry that can’t take place when you just have gases.  It’s insanely complex to study these reactions in the laboratory in controlled environments, let alone with a full, realistic suite of conditions such as you’d find in real firn.  Ice itself around the air inclusion is hardly going to just be H2O.  The snow that falls will contain dust and microbes and when you push all these together to form the ice, you’d expect some pretty interesting physico-chemical processes to be occurring.  Now anything that produces or loses CO2 is going to affect your extrapolations back to ancient ‘ambient’ CO2 levels.  If ice was that stable a reaction chamber, we’d use it to house CO2 gas standards.  But we don’t, we use expensive, high-quality, laser-polished, passivated stainless steel.

So to get a CO2 concentration that bears any likely reality to an ambient historical level in the atmosphere you need to understand and eliminate or account for:

·         The relationship between depth and age of ice (influenced by changes and nuances in the dynamics of the ice sheet and by changes in precipitation rates).

·         Physicochemical processes in the firn, and in the ice throughout the whole process from capture of ancient ambient air, through drilling, transport and analysis.

I’m being lazy, and can’t be bothered to try and understand these myself, or critique the literature adequately (partly because I know how hard it would be to do).  But my hunch is that it is for these reasons that you don’t hear that much about ice cores now.  But you do still see graphs os ‘CO2′ stretching back thousands of years – with no error bars.  Only a scientist would probably wonder about the data quality, and of those scientists almost all will be happy to trust the authors did due diligence.  Of course it would be relatively simple to do the first stage of validation experiments.  You go to somewhere where snow accumulation rates are high so you get ice core inclusions from the recent past that overlap with direct measurements.  You apply your usual (openly published!) retrieval algorithm and you see whether your ice core data agrees with the ambient measurements within the combined errors of both techniques.  If that test fails that you publish and all ice core data needs to be re-evaluated.  Assuming it passes, you’re still not out of the woods because you need to figure out how to evaluate the effects of physico-chemical processes over thousands of years, and you’ve only got 3 years for your PhD…

OK, maybe I’ve won you over and you are rightly doubtful of CO2 data from ice cores (as you should always be of data!), but you may have missed the significance of ice core data.  They are the only major reason why the elevated current CO2 levels are a cause for concern.  You have to believe that current CO2 levels are higher than they have been for thousands of years to have the spectre of a runaway greenhouse effect and tipping points.  If the ice core data is biased, then it is possible that CO2 levels today are similar to what they were in other warm periods and the mandate for drastic, immediate action is gone.  That is why the onus must be on those who would use ice core data indirectly to invoke that drastic, immediate action, to justify their methods.  Impeccably.

No doubt the politicians will keep pushing

This is the Hadley data (click the picture for a larger image). Note that 1998 is the warmest year so far. The acid test of a theory is its capacity to predict: on these grounds the IPCC computer models of a few years back are just hopeless. This in turn suggests that the anthropogenic link remains wholly speculative.

[BTW the climate change enthusiasts often quote the NASA data. This does show a rise, but NASA have been caught with their hands in the till - they have "adjusted" data they published a few years back in a way that magnifies the upward trend]

On the “preserving” – agreed – should we start a blog? With us and invitees as posters?