Reducing Uncertainty in Paleoclimate With The Kalman Filter by Anna Black ’19, Wednesday, April 10, 1:10 – 1:50 pm, Stetson Court Classroom 105, Statistics Colloquium
Abstract: Ever notice how weather forecasts 10-days out never seem to come to fruition, but near-term forecasts are relatively stable and accurate? Ever wonder why? Like most things, the answer has to do with statistics and linear algebra, in a technique called Kalman Filtering. The Kalman Filter is a tool for estimating the true state of a system, given observations and predictions about the system’s state. Crucial is the fact that both observations and predictions will have uncertainties attached to them, and the Kalman filter weighs information with less uncertainty more heavily. This Kalman Filtered based integration of observations and theory is used widely, such as in weather forecasting, navigation, and econometrics. In this talk, I focus on work that’s currently being done using the Kalman Filter to understand the Earth’s Past climate, “paleoclimate,” by combining climate models proxy records of the Earth’s climate.