Online climate change projections report 2.2 Natural variability
Climate, at a global scale and even more at a local scale, can vary substantially from one period (for example, a decade or more) to the next, even in the absence of any human influences. This natural variability of the earth’s climate has two causes. The first, natural internal variability, arises from the chaotic nature of the climate system, ranging from individual storms which affect our regional weather to large scale variations over periods of seasons to years. Variability of the latter type results mainly from interactions between ocean and atmosphere, resulting in phenomena such as El Niño. Natural internal variability will continue in future, and be superimposed on longer-term changes due to man’s activities. If in a specific future period internal variability happens to act in the same direction as man-made change then the overall change will be that much bigger; if it acts in the opposite direction, the overall change will be that much smaller. Climate models provide realistic simulations of a number of key aspects of natural internal variability in the observed climate (see Annex 3). By running the climate model many times with different initial conditions (a so-called initial condition ensemble) we can estimate the statistical nature of this natural variability on a range of space and time scales, and hence quantify the consequent uncertainty in projections.
Global temperatures projected from a three-member initial condition ensemble, all using the same emissions scenario, are shown in Figure 2.1. It can be seen that, although each experiment shows the same general warming, individual years can be quite different, due to the effect of natural internal variability. If we look at changes at a smaller scale, for example those of winter precipitation over England and Wales (Figure 2.2) we see that, although the three projections show similar upward trends of about 20% through the century, they are very different from year to year and even decade to decade. A common way of reducing the effect of uncertainty due to natural variability on the projections is to average changes over a 30 year period, as we did in the UKCIP02 scenarios (and do again in UKCP09). But even this still allows large differences in patterns of change, as can be seen from Figure 2.3; for example over Birmingham where two of the model experiments project approximately 30% increases, but the other projects just over 10%. The uncertainty due to projected natural internal variability is included in the overall uncertainty quantified in UKCP09.
There are some exciting new developments in forecasting natural internal changes in climate over the next decade, suggesting that some details of natural variability may be predictable over the next 30 years with some skill (Smith et al 2007, Keenlyside et al 2008). (We use the term skill to mean that such techniques, in which observations are used to further determine the initial state of the climate model, produce a narrower range of uncertainty than one would get in the absence of using the observations). Such techniques are still experimental, showing some promise up to a decade or so ahead with predictability beyond that yet to be tested; hence they are not used in UKCP09.
Climate can also vary due to natural external factors (that is, external to the climate system), the main ones being changes in solar radiation and in aerosol (small particles) from volcanoes. The sun is the driving force for the earth’s climate so any change in it has the potential to change climate, and indeed we estimate that the rise in global temperatures in the early part of the 20th century may have been partly due to a rise in the amount of energy reaching us from the sun over that period (Stott et al. 2003). However, because solar radiation has been relatively constant over the past few decades (apart from changes on the regular 11-year cycle which are relatively small and are largely smoothed by the inertia of the climate system) we do not attribute recent climate change over this recent period to this factor. Because we cannot forecast with any useable accuracy how the solar radiation will vary in the future, we cannot formally build any changes due to this factor in the projections of future climate; this remains as an uncertainty. However, Stott et al. (2003) have estimated that solar radiation changes over the 20th century could have caused between 0.16ºC and 0.49ºC rise in global temperatures. On the assumption that solar radiation changes over the coming century will be no greater than those in the last, although they could be in either direction, then changes in global temperature due to this factor are unlikely to be greater than ± 0.5 ºC. (Gareth Jones, pers comm.)
If volcanoes are energetic enough to inject gas into the stratosphere, then the resulting aerosol can remain there for a few years and gradually spread across the globe. Because solar radiation will be reflected back from this aerosol before it can warm the earth, it will have a cooling effect on climate at the surface. The eruption of Mt Pinatubo in the Philippines in 1991 caused global temperatures to drop by about 0.3ºC over the following year or two, taking 3–4 years to recover — and this observed effect has been quite well replicated by climate models (Hansen et al, 1996). More energetic volcanoes have an even greater cooling effect. Again, because we do not know the future course of volcanic activity, we have no meaningful way of predicting their effects on climate — apart from being aware that cooling events lasting a few years could occur at any time.