Mathematics - Colloquium
Friday, September 18, 2009
11:00 AM-12:00 PM
ABSTRACT: The atmospheric and oceanic systems are chaotic in nature which, in the context of environmental forecasting, means that small initialization errors will lead to increasing large divergence of model forecasts with respect to true trajectory. That is why weather forecasting is a challenge and the reliability of weather forecasts goes down the farther out in time we look. To reduce model initialization errors the scientific and governmental communities have embarked on unprecedented efforts to increase observing networks, launch Earth-observing satellites, and openly share data in real-time and internationally. We have transitioned from a data-starved state to becoming overwhelmed with observations. How do we ingest the massive volume of observations into models? Both observations and models have errors. Can the two random variables be combined in a Bayesian probabilistic framework? Given the volume of data (millions counts per day) and model sizes (millions of state variables), is this computationally feasible? How do chaotic models respond to observations that d not lie on any possible model trajectory? Everyone complains about the weather, but no-one does anything about it (attributed to Mark Twain). When it comes to weather forecasts however, much has been done about it.
Suggested Audiences: Adult, College
E-mail:
ma-chair@wpi.edu
Last Modified: August 25, 2009 at 3:56 PM