Eli was wandering through AGU when he came across a poster which insisted that all climate change is just a collection of jerks. This has now, in more impenetrable form (according to the author) been accepted for publication in JGR, and you know what editors are. Roger Jones proposes that continuum models just do not fit the data at all scales and intervals, that stuff changes suddenly.
Past climate change has often been rapid. Although future climate change could also be rapid, it is usually communicated as a trend.Jones has looked at a number of changes, including air and sea temperatures at local to global scales, precipitation, and more.
Climate projections developed from coupled atmospheric ocean models are interpreted through a signal to noise model. This model interprets the anthropogenic signal in climate variable (temperature, rainfall, etc.) as a smooth curve and variability as noise around that curve..
For long time series of annual data, annual to decadal variability is interpreted as being of natural origin.
Were these changes natural variability superimposed on a long term trend or were they a non-linear response to anthropogenic forcing
The resulting investigation has concluded that anthropogenic climate change is a non-linear process. Rather than following the signal-to-noise model, climate change follows a step and trend process.
The bunnies thought about this, and the useful analogy is earthquakes. Stress (e.g. forcing) builds steadily, but change happens suddenly. Averaging, smooths change to a fair-thee-well, continuum models do the same. The uncertainty monster is not anyone's friend.