 |
|
 |
 |
 |
|
Navigation
Tomislava Vukicevic, Atmospheric and Oceanic Sciences, University of Colorado, Boulder
/ Home / Events / Tomislava Vukicevic, Atmospheric and Oceanic Sciences, University of Colorado, Boulder
|
Tomislava Vukicevic, Atmospheric and Oceanic Sciences, University of Colorado, Boulder
A new look at data assimilation and inversion problems in atmospheric sciences. Data assimilation and inversion problems are involved in almost every aspect of quantitative analysis in atmospheric sciences from observing by indirect measurements to prediction and projections by numerical models. Contemporary literature on common data assimilation and inversion techniques typically refers to theoretical basis of the techniques as a straightforward generalization of the Bayesian rule for conditional probabilities when applied under assumptions of errorless linear model and Gaussian statistics. The data assimilation and inversions are however, more often than not done with complex nonlinear models and observations for which these assumptions are not necessarily valid. To help understand impact of model nonlinearities, Gaussian statistics and modeling and observation errors a new approach is used which is based on a generalized formulation of the statistical inverse problem theory that does not make explicit use of the Bayesian rule. Based on the new approach an analysis of data assimilation and inversion solutions was done by numerical evaluation of the associated probability density functions on examples of two relatively simple but representative dynamical models of atmospheric processes. Relationship between properties of common data assimilation and inversion techniques and the generalized solutions will be discussed with an outlook at furthering benefits from inverse methodology in atmospheric sciences. April 3, 2008 3:30 pm - 4:29 pm 2246 SRB Auditorium |
© 2003 - 2008 Univ. of Michigan