Tuesday,
November 22, 2005, 12:15 p.m. GWC 216
Data Assimilation, Inversion and Statistical Hypothesis Testing
Abstract
What do these three areas have in common? Data assimilation
is typically used to improve model predictions with observed
data. Inversion is commonly used to infer the parameters
in these models given the observed data. Data assimilation and
inversion are virtually equivalent, they just tend to emphasize
different aspects, i.e. data assimilation emphasizes improving
model prediction and inversion emphasizes parameter estimation.
However, data assimilation is becoming more commonly used for
parameter estimation. To bring these three concepts together I
will show how data assimilation or inverting for parameter
estimates can be viewed as statistical hypothesis testing. This
viewpoint allows us to use some new ideas to improve parameter
estimates.
For further information please contact:
mittelmann@asu.edu
