pyesmda.FixedLocalization

class pyesmda.FixedLocalization(correlation_matrix: ndarray[tuple[Any, ...], dtype[float64]] | spmatrix | None = None)

Fixed localization strategy.

Variables:

correlation_matrix (Optional[Union[sp.sparse.sparray, NDArrayFloat]]) – Correlation matrix based on spatial and temporal distances between observations and \(\rho_{DD}\). It is used to localize the autocovariance matrix of predicted data by applying an elementwise multiplication by this matrix. Expected dimensions are (\(N_{\mathrm{obs}}\), \(N_{\mathrm{obs}}\)).

__init__(correlation_matrix: ndarray[tuple[Any, ...], dtype[float64]] | spmatrix | None = None) → None

Initialize the instance.

Parameters:

correlation_matrix (Optional[Union[NDArrayFloat, spmatrix]]) – Correlation matrix based on spatial/temporal distances between observations/parameters \(\rho_{DD}\) or \(\rho_{MD}\). It is used to localize the empirical cross-covariance matrices by applying an elementwise multiplication by this matrix. Expected dimensions are (\(N_{\mathrm{obs}}\), \(N_{\mathrm{obs}}\)) for \(\rho_{DD}\) and (\(N_{m}\), \(N_{\mathrm{obs}}\)) for \(\rho_{DD}\). It None, no localization is performed. The default is None.

Properties

Methods

check_localization_shape	Check if
localize	Apply the localization to the covariance matrix.
localize_multi_dot	_summary_