If climatic disasters such as severe droughts can be predicted a season or two in advance, their impacts can be considerably mitigated. With advances in numerical modeling capabilities, it has become possible to make such long-term forecasts, although their skill is quite modest when compared to short-term weather forecasts. This chapter reports on the current status of “dynamical climate prediction” (i.e., prediction of climate variations on seasonal-to-decadal time scales using comprehensive computer models of the climate system). Dynamical climate predictions are now competitive with empirical predictions made using statistical models based on historical data. Dynamical prediction skill is currently limited by errors in the formulation of numerical climate models, as well as by errors in the initial conditions. Increases in computational power, better model formulations, and new ocean observing systems are expected to lead to improved prediction skill in the future.

Introduction

Predictions, prognostications, and prophecies have always fascinated the human mind over the ages. In the early days of civilization, the ability to predict often connoted power and religious authority. Regardless of whether the predictions were right or wrong, they still affected people's lives. Predictions have lost much of their magical aura in the modern age, but they still play an important role in our lives.