Abstract
How do you start a weather model in a million dimensional space? Or better: Where? Our models make contact with reality at two points: first, in data assimilation, when we use observations to help us choose initial conditions; second, in constructing forecasts to move back from model-land to the real world. Concepts from the theory of nonlinear dynamical systems suggest a new approach to data assimilation in chaotic systems which are shown to work well. It also implies that in operational weather forecasting, data assimilation lacks a mathematically well-defined aim. This suggests that while we can extract quantitative guidance from weather models (but probably not probabilities), we might better aim to extract understanding and insight from climate models (but not numbers). Where numbers are given, bolstering statistical good practice in climate science for the next IPCC assessment might increase both its utility and its impacts.
Biography
Professor Leonard Smith was raised in Florida (USA), receiving a B.S. (Honours) in Physics Mathematics and Computer Science from the University of Florida and a PhD in Physics from Columbia University. He has been a Senior Research Fellow in mathematics at Pembroke College Oxford since 1992. In 2004, he became a Professor of Statistics (research) at the London School of Economics, where he is Director of the Centre for the Analysis of Time Series (CATS); he also directs programmes within the LSE’s Grantham Research Institute. His research focuses on nonlinear dynamical systems, predictability, the role of probability in decision support, and the implications uncertainty, ambiguity and ignorance hold when relating mathematical results to reality. His 2007 book, A Very Short Introduction to Chaos, is one of the best selling “technical” volumes in the OUP series.
He is a member of the American Statistical Association’s Advisory Committee on Climate Change Policy (ACCCP) and has worked on an Expert Team of the WMO on forecast verification. As a Selby Fellow of the Australian Academy of Science he gave public lectures across Australia. In 2003, the Royal Meteorological Society awarded him the Fitzroy Prize in recognition of his contributions to applied meteorology.
Note on attendance:
Grantham Seminars are open to all but please note that we ask all external visitors to email Gosia Gayer to register.
Directions to the Institute are available on our site.