endas.localization.GenericStateSpace1d

class endas.localization.GenericStateSpace1d(n)

Bases: endas.localization.domain.StateSpacePartitioning

Analysis localization assuming generic state space.

This a generic form of domain-based analysis localization where each state space variable is assigned its own local domain. The local analysis update is therefore performed for each state variable individually. Under this scheme, the indexes of state vector elements also act as their respective “coordinates”. Furthermore, observations are assumed to be directly related to state vector elements and their coordinates are also expressed as the indexes of the corresponding state vector elements.

Parameters:n – The number of elements in the state vector

Note

Please note that this localization scheme is rather arbitrary and is unlikely to be suitable for any real geophysical system. This is especially true if non-trivial observation operators are used or in case of multivariate data assimilation. The primarily purpose of GenericStateSpace1d is for testing analysis localization on synthetic problems.

The localization scheme is not suitable for large state spaces as the number of local domains is equal to the size of the state vector. This will make the localized analysis very slow for large n.