Second Order Methods for Error Propagation in Variational Data Assimilation

Abstract : This chapter discusses the use of second-order methods for estimating error propagation in variational data assimilation. The basic variational approach to data assimilation exhibits the optimality system: it can be considered as a generalized model containing all the available information. To estimate the impact of errors due to the parameters of the model and/or to the observations, it is necessary to consider second-order properties. The variational approach can be used to estimate the propagation of uncertainties in the analysis. Two basic cases are considered. In the deterministic framework, the uncertainty is a virtual and deterministic perturbation on the model parameters, whose impact on some criterion is to be found. In the stochastic framework, the uncertainty is a random variable transported by the model as such. The output is a stochastic perturbation on the outputs of the analysis, for which it is necessary to determine its probabilistic characteristics.
Document type :
Book sections
Complete list of metadatas

https://hal.inria.fr/hal-01103169
Contributor : Eugene Kazantsev <>
Submitted on : Wednesday, January 14, 2015 - 10:57:17 AM
Last modification on : Monday, February 4, 2019 - 3:30:29 PM

Identifiers

Collections

Citation

François-Xavier Le Dimet, Igor Gejadze, Victor P. Shutyaev. Second Order Methods for Error Propagation in Variational Data Assimilation. Eric Blayo; Marc Bocquet; Emmanuel Cosme; F. Cugliandolo Leticia. Lecture notes of Les Houches summer school, Oxford University Press, pp.576, 2014, Data assimilation for Geosciences, 9780198723844. ⟨10.1093/acprof:oso/9780198723844.003.0014⟩. ⟨hal-01103169⟩

Share

Metrics

Record views

302