Event Details

Professor Peter Young will be giving the first of this year's DSI Distinguished Seminar Series with a talk on the subject of modelling and forecasting global temperature.

Abstract

The best known global climate models are the very large and complex Atmosphere-Ocean General Circulation Models (AOGCMs). In this talk, Professor Young will show that these very large models are not the best basis for forecasting globally averaged surface temperature. In particular, they do not capture efficiently the information on the 'dominant modes' of dynamic behaviour required for short to medium term forecasting applications. Indeed, he will show that superior forecasts can be obtained using very much smaller models that can be identified directly from the globally averaged data using my Data-Based Mechanistic (DBM) approach to modelling stochastic, dynamic systems. The resulting DBM differential equation model relates the globally averaged measure of Total Radiative Forcing (RTF) to the surface temperature, as measured by the Global Temperature Anomaly (GTA). In contrast to the AOGCM’s, this produces a good prediction of the `levelling' in the GTA over the period 2000 to 2011. This prediction derives in part from the effects of a 40-50 year period, a quasi-cyclical component that is not present at all in the AOGCM projections and is shown to be correlated with the Atlantic Multidecadal Oscillation (AMO) index. Interestingly, the DBM model also suggests that the Equilibrium Climate Sensitivity parameter, used by climate scientists as a measure of long-term global temperature change, is quite a lot smaller than that suggested previously, something that has been confirmed by research reported more recently in the journal Nature. Finally, at a more philosophical level, Professor Young will argue that, when dealing with natural systems such as the global climate, the ‘hypothetico-inductive’ method of DBM modelling has advantages over the well known 'hypothetico-deductive’ approach suggested by Karl Popper and favoured by climate scientists.

An associated paper, published last year in the International Journal of Forecasting, is available from: https://www.dropbox.com/s/1jsafkxh452a5jt/PCY_IJF_2017.pdf?dl=0