I got into another slightly silly discussion yesterday about climate models, which I tried to resolve by directing the other person to a new paper by Julia Hargreaves and James Annan called Can we trust climate models?. This is a paper I quite like because it both presents the aspects of climate models that we might regard as robust and the aspects that are less certain.
For example, the paper says,
It seems that genuinely useful climate forecasting on the multiannual to decadal timescale may be still some way away at this time. Thus it is clear that the models can currently only be relied upon for a broad picture of future climate changes.
Seems reasonable to me. Individual models may be able to represent decadal variability, but collectively models are still unable to reliably predict what might be expected in the coming years or decade. I would actually argue that this is one reason why claiming that climate models have failed because they didn’t predict the “hiatus” is wrong. They were never really capable of predicting such a hiatus and so suggesting that they’re wrong because they didn’t do something they were not capable of doing seems a little silly.
The paper also says,
On the regional scale there are, however, substantial disagreements in magnitude and pattern of temperature anomalies both between models and data, and also within the model ensemble. Therefore, we cannot expect precise predictions from current climate models.
In fact, models are very far from being perfect. They struggle to generate robust simulations of recent climate changes on regional scales, even when run at the highest resolutions available.
So, models don’t perform particularly well at the regional scale. Not that surprising. Even the highest resolutions are probably still unable to properly resolve these smaller scales and I assume that some of the relevant physics is parametrised, rather than self-consistently evolved by the models. I would add, however, that it’s unlikely that models have no value at these regional scale. I can go back 20 years and read papers in my field that present results from simulations that used resolutions that we would never consider using today. Today, we might understand such systems in much more detail than we did 20 years ago, but the results from these early simulations were not valueless. It’s a process of evolution; we don’t go from “wrong, ignore” to “right, accept”.
The paper does, however, say
Probably the most iconic and influential result arising from climate models is the prediction that, dependent on the rate of increase of CO2 emissions, global and annual mean temperature will rise by around 2–4∘C over the 21st century. We argue that this result is indeed credible, as are the supplementary predictions that the land will on average warm by around 50% more than the oceans, high latitudes more than the tropics, and that the hydrological cycle will generally intensify.
So, the overall warming, the variation between the land and oceans, and changes to the hydrological cycle are well represented by the models. These is probably because these largely represent the bits of physics (radiative transfer, heat contents of the different components of the climate system, water cycle) that we understand well.
So, some aspects of climate models are robust and can probably be trusted (overall warming, water cycle) others are less robust and should be treated with caution (regional and decadal predictions). What does this really mean? As far as I’m concerned it just means we have to be aware of these issues, take them into account, and work with what we have. As the paper itself says
Far more problematic, is that we are unwilling to wait 100 years before learning about climate models, and cannot wait before making today’s decisions.
I should add, however, that climate models are not the only evidence for climate change and so just because we can’t trust all aspects of today’s climate models doesn’t really mean that we have no reliable evidence for climate change or that we should simply decide to wait until we have climate models that are better at the representing the regional and decadal scales.