Can We Predict Climate Fluctuations?

The climatic state is defined in terms of the statistics of the complete set of atmospheric, hydrospheric, and cryospheric variables over a specified time interval in a specified domain of the earth-atmosphere system. It is established that a certain amount of climatic variability which arises from day-to-day weather fluctuations is unpredictable. The nature and magnitude of this inherent variability is discussed. Causes of potentially predictable variability are outlined; these include changes in both the external and internal climate system. There are basically three methods of forecasting this variability: wholly statistical procedures, physical-empirical methods and wholly dynamical methods. The advantages and disadvantages of each method are discussed, with particular emphasis on the use of numerical models of the atmospheric circulation. It is concluded that the potential of statistical methods is limited because of lack of data and because ultimately forecasts should be based on an understanding of the system. Climate models are useful tools for understanding the climate system but can not, at the present time, be used for predicting climate fluctuations on an interannual time scale. Physical-empirical methods already show some success but further work is required before we can confidently make reliable interannual climate predictions