Predictability

Abstract

A group of junior faculty members and UCAR junior scientists convened in Boulder, CO on June 16-18, 2003 for discussion on future scientific directions. This report summarizes the goals and products of one of the three foci selected for special consideration: predictability. About 15 people, representing physical, mathematical, and biological sciences, were present for round-table discussions. The discussion sought common interpretation of the predictability problem, points of generalization, identification of major hurdles, and potential approaches to their solution. The diverse background of the participants generated a wide-ranging discussion. The participants addressed predictability generally, while supplying specific examples from their own areas of expertise. Recurring themes included the relationship between models and initial conditions, the importance of definitions and the choice of a norm for evaluation, and generalization across systems and disciplines. The group explored potential avenues for generalization through interdisciplinary networking. Short- and long-range challenges were identified related to probabilistic state estimation, verifying predictions and understanding error, and dealing with nonlinearity. In this essay we expand on these themes and challenges, and describe possible future research objectives.