 |
| Titel |
Precision variational approximations in statistical data assimilation |
| VerfasserIn |
J. Ye, N. Kadakia, P. J. Rozdeba, H. D. I. Abarbanel, J. C. Quinn |
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| ISSN |
2198-5634
|
| Digitales Dokument |
URL |
| Erschienen |
In: Nonlinear Processes in Geophysics Discussions ; 1, no. 2 ; Nr. 1, no. 2 (2014-10-10), S.1603-1620 |
| Datensatznummer |
250115127
|
| Publikation (Nr.) |
 copernicus.org/npgd-1-1603-2014.pdf |
|
|
|
|
|
| Zusammenfassung |
| Data assimilation transfers information from observations of
a complex system to physically-based system models with state
variables x(t). Typically, the
observations are noisy, the model has errors, and the initial state
of the model is uncertain, so the data assimilation is
statistical. One can thus ask questions about expected values of
functions ⟨G(X)⟩ on the path X =
{x(t0), ..., x(tm)} of the model as it moves through an
observation window where measurements are made at times
{t0, ..., tm}. The probability distribution on the path
P(X) = exp[−A0(X)] determines these expected values. Variational
methods seeking extrema of the "action" A0(X), widely known as
4DVar (Talagrand and Courtier, 1987; Evensen, 2009),, are widespread for estimating
⟨G(X) ⟩ in many fields of science. In a path integral
formulation of statistical data assimilation, we consider
variational approximations in a standard realization of the action
where measurement and model errors are Gaussian. We (a) discuss an
annealing method for locating the path X0 giving a consistent
global minimum of the action A0(X0), (b) consider the explicit
role of the number of measurements at each measurement time in
determining A0(X0), and (c) identify a parameter regime for the
scale of model errors which allows X0 to give a precise estimate
of ⟨G(X0)⟩ with computable, small higher order
corrections. |
| |
|
| Teil von |
|
|
|
|
|
|
|
|