pyPDAF.PDAFomi.observation_localization_weights

pyPDAF.PDAFomi.observation_localization_weights(i_obs: int, ncols: int, a_l: np.ndarray, dim_obs_l: int, verbose: int) → np.ndarray

Returns a vector of observation localisation weights.

The weights are based on specifications given by localisation setups and observation coordinates in OMI. This function is used in the case of non-diagonal observation error covariance matrix where one has to perform localisation in user-supplied functions, e.g., pyPDAF.c__prodrinva_pdaf() or pyPDAF.c__likelihood_l_pdaf().

Here, a_l is typically the input array in pyPDAF.c__prodrinva_pdaf().

Parameters:

  • i_obs (int) – index into observation arrays

  • ncols (int) – Rank of initial covariance matrix

  • a_l (ndarray[np.float64, ndim=2]) – Input matrix (thisobs_l%dim_obs_l, ncols) Array shape: (:, :)

  • dim_obs_l (int) – Dimension of local observation vector of the i_obs-th observation type

  • verbose (int) – Verbosity flag

Returns:

weight – > Localization weights Array shape: (thisobs_l(i_obs)%dim_obs_l)

Return type:

ndarray[np.float64, ndim=1]