Variance Estimates in Seasonal Climate Forecasting and Related Food Reserve Requirements

Abstract

A general assessment of the potential reduction in variance of predicted climatological statistics was made on the supposition that long-term seasonal-to-annual climatic averages are causally predictable. The residual variance, stemming from short-term unpredictable variability (the ?weather?), was found to be of the order of magnitude comparable to observed interannual changes when estimated on a local basis. When the geographic area was extended to global dimensions, the latter became dominant, indicating on this scale the potential for a significant reduction in climatic mean variance if a seasonal climatic prediction model were available. The relevance of these results to food reserve policy and to the directions of climatological research is discussed.