CLIVAR Science

Global Warming in a Chaotic Climate - A Simple Conceptual Model

Our climate is chaotic. As a result, the internal variations of climate are irregular and weather patterns never repeat themselves exactly. Moreover, because of chaos, predicting the detailed evolution of weather over long periods of time is effectively impossible; the tiniest error in specifying the initial state in a weather forecast will eventually cause the forecast to deviate completely from reality.

How, therefore, are we to understand the problem of anthropogenic climate change in a chaotic climate? In particular, does the problem of chaos make the prediction of climate change effectively impossible?

The animation shows the familiar "Executive Decision Maker" - a metal pendulum swinging over four magnets. When the pendulum is swinging, the system is effectively chaotic in that a time series of "magnetic regimes" over which the pendulum oscillates, appears quite irregular; and if the pendulum were started from a slightly different position, a very different time series of magnetic regimes could ensue. Imagine this system as an analogy for climate. For the sake of argument, suppose the magnets correspond to different climatic regimes: the yellow magnet to an anomalously mild winter, the white magnet to an especially cold winter, the red and blue magnets to other winter states having more-or-less normal temperatures.

If the initial state of the pendulum corresponds to the weather conditions on Jan 1 1800, then the time-series of pendulum position corresponds to a plausible 19th century evolution of climate (illustrative, let's say, of pre-industrial times). Because of chaos, it would be impossible to predict the precise sequence of winter regimes in the 19th century, given some estimate of initial conditions for Jan 1 1800. Nevertheless, one can say that there is a 25% probability of occurrence, for any particular 19th century winter season, of any one of the four winter climatic regimes.

The second animation shows the same pendulum system, but with a wedge moved in underneath the base. The wedge represents an external perturbation to the system, here the perturbation associated with anthropogenic increases in carbon dioxide. With the wedge firmly in place - representing, let's say, late 21st century levels of carbon dioxide - let's see how the system evolves. The motion is still chaotic, so that late 21st century weather variations are still irregular and therefore impossible to predict in detail. However, one can clearly see that the probabilities of occurrence of the four winter regimes are no longer equal - the mild winter is now significantly more likely that the cold winter, even though the system does make occasional unpredictable excursions to the cold winter regime.

These animations [courtesy of Rob Hine (ECMWF)] illustrate a number of climate issues which have caused conceptual difficulties from time to time:

Even though climate is chaotic, with weather states impossible to predict in detail more than a few days ahead, there is a predictable impact of anthropogenic forcing on the probability of occurrence of the naturally-occurring climatic regimes. This lies at the heart of the CLIVAR perspective on climate change - how anthropogenic forcing will affect the natural modes of climate variability.

In our chaotic climate, it is impossible (indeed meaningless) to try to attribute a specific (eg severe) weather event to anthropogenic global warming. Hence, it is a false dichotomy to suppose that some recently-occurring drought or flood is either on the one hand caused by global warming, or on the other hand is merely due to natural climate variability.

Rather, the correct way to address such an issue is to ask instead whether anthropogenic climate change will increase or decrease the probability of occurrence of the type of drought or flood which we (or journalists pursuing some weather story provoked by a recent drought or flood) are interested. Such probabilities can be obtained, for example, from the JSC/CLIVAR Working Group on Climate Modelling's multi-model ensemble, made for the IPCC fourth assessment report.

In a chaotic climate, one cannot expect the time-series of global temperature to increase monotonically under the impact of anthropogenic climate change. Hence, for example, global mean temperatures were especially warm in 1998 because of the occurrence of a substantial El-Niño event. By the bullet above, it is meaningless to attribute the 1998 El-Niño event to global warming. Only by looking over long enough periods of time can one see the trend in global mean temperature due to anthropogenic climate change, above the "noise" of climatic variability.