Predictability, uncertainty and decision making: a unified perspective to build a bridge from weather to climate

In this essay, the common thread of limits of predictability and uncertainty that permeate across weather and climate prediction and projections is discussed in the context of developing a strategy for 'seamless' communication and utilization of uncertain information in decision making. In understanding why uncertainty is an unavoidable trait of predictions in the first place, a useful concept is the separation of the Earth System (ES) into internal and external components. This separation allows one to first, recognize that for prediction at all time-scales, the inherent source for limits on predictability is due to the divergence of forecasts from a cloud of initial conditions, and second, thereby recognize that the fundamental source of uncertainty (or unpredictability) is limited by our ability to specify initial conditions for the internal component with perfect accuracy. The unavoidability of uncertainty in predictions, and accepting this fact could be advantageous in the ongoing discussions on how to communicate climate projections and the associated uncertainties by learning from the knowledge base that exists for communicating similar information on weather and seasonal predictions that are generated on a much more frequent basis. Similarly, decision-support systems for developing adaptation and mitigation strategies can use predictions on shorter range as a test-bed to hone their strategies to incorporate predictive uncertainty when dealing with longer-range projections. By practicing the use of decision making tools and the incorporation of uncertain predictions on weather and seasonal time scale, decision makers can improve their level of comfort in accepting uncertainty inherent in longer range predictions and projections on a much less infrequent basis. In this paradigm, evolving strategy for seamless predictions can be blended with a strategy for seamless communication of uncertain information and also with seamless application of decision support systems.