Thursday, May 8, 2014

Scientific models are not actually some obscure mystery

Just a quick one on science communication.

Via Pharyngula I learned of a piece in the Guardian, The top ten global warming 'skeptic' arguments answered. Dana Nuccitelli replies to a number of claims or questions made by one of the few climate scientists who still doubt human-caused climate change.

Some of those arguments are actually very twee, and the rebuttals are accordingly amusing, as in the case where the 'skeptic' asks "how do scientists expect to be taken seriously when their theory is supported by both floods AND droughts?" I guess if faced with a breakdown of cargo transportation he would be just as confused why there can be food shortages in one area but undelivered produce rotting in the other - surely it has to be either-or, amirite? Guess stuff like that is complicated.

But as funny as it is, there was one minor point that annoyed me a bit, and that is the answer to the claim that we cannot trust climate models, which went like this:
Climate models have accurately reproduced the past, but let's put them aside for a moment. We don't need climate models to project future global warming. We know from past climate change events the planet will warm between about 1.5 and 4.5°C from the increased greenhouse effect of a doubling of atmospheric carbon dioxide
Excuse me if I am wrong (because I am only a scientist), but it seems to me that the expectation of 1.5 to 4.5°C in warming for each doubling of CO2 is, well, a model. It is a very simple model, but it is a model, because it is a mathematical description of the behavior of a system.

What we see in the Guardian piece is unfortunately rather typical of the public discussion of science. Science is treated as some kind of mystical process happening in the ivory tower but inaccessible to Jane Doe and Joe McAverage. In reality, science is nothing but a formalized version of what everybody does in their daily lives as long as they are unencumbered by religious or ideological blinkers: Accepting the simplest explanation over unnecessarily convoluted ones, testing different ideas against observable reality, demanding exceptional proof for exceptional claims, and so on.

And yes, modelling. Intuitively, we all do it every day. If you are a teacher or lecturer, you may, for example, know from experience that you can grade three student course papers of a certain type in one hour. You would then perhaps assume that you can achieve the generalized outcome of
student papers graded = 3 x time invested in hours.
That is a simple model. In reality, you will find that you get tired and less efficient if you keep on going for too long without breaks. You may then want to adjust this formula to take into account that the factor of papers per hour diminishes over time, so that you may only get eight done in three hours. This would be a slightly more sophisticated model, but the principle is still the same. You put the hours into the model and you get the expected number of graded papers out.

Likewise, really sophisticated computer modelling of climate change probably involves many factors. I am not an expert in that area, but I would expect that CO2, water vapor, cloud formation, ocean temperatures, methane, and planetary albedo all play a role, plus many others. And then there might be positive and negative feed-backs in the model; for example, higher temperature may lead to more water vapor as more water evaporates, and because water itself is a greenhouse gas it may lead to even more warming, etc.

But all that complexity of the sophisticated models does not mean that "ca. 3°C per doubling of CO2" is not also a model. Heck, even the position "there is no significant influence of CO2 on climate" is a climate model, so the position that no models can be trusted is self-refuting!

And this is then what is frustrating. When faced with the usual denialist's canard that you cannot trust any models anyway, most people immediately either grant that argument or, as in this case, change the topic after a flat assertion, so that the on-lookers will walk away with the absurd impression that modelling is indeed an obscurantist, useless egghead's plaything. Instead, we should make an effort to demystify the concept of modelling and try to explain what that word actually means.

Scientific models are really only formalized versions of what all of us constantly rely on to predict the behavior of the world around us. In fact, there is simply no way to even discuss the dynamic behavior of any system - be it a factory, a nation's economy, a human asking for a diet plan, or the global climate - without using models, at least very simplistic ones.

No comments:

Post a Comment