SD Conf. – VIII – Sterman et al – Climate Change modelling with C-ROADS

By | August 4, 2009

Notes from one of the closing plenaries at the 27th System Dynamics Conference::

Using C-ROADS to Support Analysis of International Climate Change Proposals by Andrew Jones, John Sterman, Thomas Fiddaman, Travis Franck, Elizabeth Sawin

Following on nicely from the presentation by Moxnes, in this presentation John Sterman talked about climate change, modelling, and decision making.

By way of introduction, he showed graphs of data derived from the International Panel on Climate Change (IPCC) models plotted alongside real world measurements. At the moment, we’re doing worse than the worst-case scenario predicts.

Next, he presented a study showing that when asked, 60% of MIT graduate students thought that stabilizing emissions would stabilize the level of carbon dioxide in the atmosphere, which is flat out wrong1. Stabilization can only occur when net emissions equals net removal, which means that we either have to decrease emissions, or somehow increase removal. To make things worse, a large number of policy makers and other supposed experts made the same mistake.

Having scared us a bit, Sterman moved on to talk about why climate change is such a hard policy nut to crack:

  • The science of climate change consists of large numbers of causal linkages of varying magnitudes whose dynamic interactions can only be understood through complex models. These are:
    • Far too complex for humans to reason through unaided, and so intuition about climate change is usually misleading.
    • Technically complex and require large amounts of computation to run.
    • Not transparent to policy makers, who only see results based on parameters chosen by experts whose judgement they might question.
    • Not accessible to policy makers who might otherwise experiment with policy proposals in order to gain an intuitive approximation of what’s going on.
  • Arguments about climate change incorporate statistical reasoning, another area in which human intuition is unreliable.
  • Policy proposals often rely on varying time frames, use different units, and are expressed relatively from different base figures. It’s difficult to reason about a 20% reduction from 1990 levels compared to a 25% reduction from 2004 levels, particularly given other possible parameters.

Of course, there’s political reasons for climate change being hard, too, such as the requirement for all actors to participate for a policy to truly be effective. But in the end, the biggest problem is simply education and the availability of information.

So, Sterman’s group have been working on a model they called C-ROADS, standing for Climate Rapid Overview And Decision Support. It’s an interactive System Dynamics model that captures most of the feedback loops involved in climate change and behaves similarly to the IPCC models. Since it’s interactive and relatively simple to use, it fills a gap in the discourse, allowing policy makers to do all those things I just said they couldn’t using the IPCC climate models. Furthermore, there’s an online version of the model called C-LEARN at

Having that model to play with makes it all a lot more real. Here’s a dirty secret. If you model all of the publicly available proposals for reducing emissions, by 2100, the amount of carbon in the atmosphere is 600 parts per million, and still rising. That’s a lot more than the 350 parts per million target that’s been often mentioned. Though, of course, we’re already exceeded that, so the target’s been moved to 400.

600 parts per million, when you take into account a mild amount of glacial melting and thermal expansion, means a world sea level rise along a shallow exponential curve to hit about 4 metres by 2100. Take into account the melting of the various ice sheets, non-carbon dioxide greenhouse gases, and the more hypothetical feedbacks such as the clathrate gun, and it gets much worse. Unfortunately, he didn’t talk much about these, nor are they available in the C-LEARN model.

To stabilize atmospheric carbon any time before 2100, you basically need everyone to agree to something like 80% cuts in emissions from 2004 levels by 2050. Interestingly, Europe’s already done that, Canada’s close, and if you can believe the Obama rhetoric, the US government at least wants to do that. Unfortunately, everyone needs to do this, even eventually including the African nations, who aren’t emitting much at all right now. Oh, and by the way, the New Zealand National government’s proposals are clearly inadequate, particularly when one takes into account our agricultural methane production..

Of course, even if we can manage cuts like that, we’ve already committed ourselves to about a 2 degree rise in temperature, which means at least a metre rise in sea level.


On the other hand, tools like this makes it a lot harder to obfuscate and mislead oneself about the measures need to address the problem, and from what I hear, it’s been used in training sessions with politicians in a bunch of countries, with remarkable results. So, there’s room for a little optimism, I guess.

[1] It’s not enough to stop emissions growth. If it’s not clear why that’s true, think about a bathtub. If the tap is pouring in more water than is draining, the tub will gradually fill. Same applies to carbon dioxide in the atmosphere. Basically, as described earlier, people suck at understanding dynamics.